telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1 | // |
Matthew Sloyan | ca36123 | 2023-02-16 14:50:22 +0000 | [diff] [blame] | 2 | // Copyright © 2017,2022-2023 Arm Ltd and Contributors. All rights reserved. |
David Beck | ecb56cd | 2018-09-05 12:52:57 +0100 | [diff] [blame] | 3 | // SPDX-License-Identifier: MIT |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 4 | // |
| 5 | #include "OnnxParser.hpp" |
| 6 | |
Matthew Sloyan | ac001ee | 2021-02-03 10:43:04 +0000 | [diff] [blame] | 7 | #include "armnnOnnxParser/Version.hpp" |
| 8 | |
Matthew Bentham | 39ef3e5 | 2020-01-20 10:09:09 +0000 | [diff] [blame] | 9 | #include <armnn/Descriptors.hpp> |
Narumol Prangnawarat | ac2770a | 2020-04-01 16:51:23 +0100 | [diff] [blame] | 10 | #include <armnn/utility/Assert.hpp> |
Matthew Sloyan | 589e3e8 | 2020-09-11 16:17:48 +0100 | [diff] [blame] | 11 | #include <armnn/utility/NumericCast.hpp> |
Narumol Prangnawarat | bc3bb62 | 2021-09-24 16:08:34 +0100 | [diff] [blame] | 12 | #include <ParserHelper.hpp> |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 13 | #include <VerificationHelpers.hpp> |
| 14 | |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 15 | #include <fmt/format.h> |
Aron Virginas-Tar | d4f0fea | 2019-04-09 14:08:06 +0100 | [diff] [blame] | 16 | |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 17 | #include <google/protobuf/text_format.h> |
| 18 | #include <google/protobuf/io/zero_copy_stream_impl.h> |
| 19 | |
Matthew Sloyan | ac001ee | 2021-02-03 10:43:04 +0000 | [diff] [blame] | 20 | #include <iostream> |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 21 | #include <numeric> |
Jan Eilers | 53ef795 | 2021-06-02 12:01:25 +0100 | [diff] [blame] | 22 | #include <armnnUtils/Permute.hpp> |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 23 | |
| 24 | using namespace armnn; |
| 25 | |
| 26 | namespace armnnOnnxParser |
| 27 | { |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 28 | |
| 29 | IOnnxParser::IOnnxParser() : pOnnxParserImpl(new OnnxParserImpl()) {} |
| 30 | |
| 31 | IOnnxParser::~IOnnxParser() = default; |
| 32 | |
| 33 | IOnnxParser* IOnnxParser::CreateRaw() |
| 34 | { |
| 35 | return new IOnnxParser(); |
| 36 | } |
| 37 | |
| 38 | IOnnxParserPtr IOnnxParser::Create() |
| 39 | { |
| 40 | return IOnnxParserPtr(CreateRaw(), &IOnnxParser::Destroy); |
| 41 | } |
| 42 | |
| 43 | void IOnnxParser::Destroy(IOnnxParser* parser) |
| 44 | { |
| 45 | delete parser; |
| 46 | } |
| 47 | |
| 48 | armnn::INetworkPtr IOnnxParser::CreateNetworkFromBinaryFile(const char* graphFile) |
| 49 | { |
| 50 | return pOnnxParserImpl->CreateNetworkFromBinaryFile(graphFile); |
| 51 | } |
| 52 | |
Mike Kelly | 2ae3224 | 2022-11-25 13:55:24 +0000 | [diff] [blame] | 53 | armnn::INetworkPtr IOnnxParser::CreateNetworkFromBinary(const std::vector<uint8_t>& binaryContent) |
| 54 | { |
| 55 | return pOnnxParserImpl->CreateNetworkFromBinary(binaryContent); |
| 56 | } |
| 57 | |
| 58 | armnn::INetworkPtr IOnnxParser::CreateNetworkFromBinary(const std::vector<uint8_t>& binaryContent, |
| 59 | const std::map<std::string, armnn::TensorShape>& inputShapes) |
| 60 | { |
| 61 | return pOnnxParserImpl->CreateNetworkFromBinary(binaryContent, inputShapes); |
| 62 | } |
| 63 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 64 | armnn::INetworkPtr IOnnxParser::CreateNetworkFromTextFile(const char* graphFile) |
| 65 | { |
| 66 | return pOnnxParserImpl->CreateNetworkFromTextFile(graphFile); |
| 67 | } |
| 68 | |
| 69 | armnn::INetworkPtr IOnnxParser::CreateNetworkFromString(const std::string& protoText) |
| 70 | { |
| 71 | return pOnnxParserImpl->CreateNetworkFromString(protoText); |
| 72 | } |
| 73 | |
Narumol Prangnawarat | 1b11f32 | 2021-10-13 11:44:50 +0100 | [diff] [blame] | 74 | armnn::INetworkPtr IOnnxParser::CreateNetworkFromBinaryFile( |
| 75 | const char* graphFile, |
| 76 | const std::map<std::string, armnn::TensorShape>& inputShapes) |
| 77 | { |
| 78 | return pOnnxParserImpl->CreateNetworkFromBinaryFile(graphFile, inputShapes); |
| 79 | } |
| 80 | |
| 81 | armnn::INetworkPtr IOnnxParser::CreateNetworkFromTextFile(const char* graphFile, |
| 82 | const std::map<std::string, armnn::TensorShape>& inputShapes) |
| 83 | { |
| 84 | return pOnnxParserImpl->CreateNetworkFromTextFile(graphFile, inputShapes); |
| 85 | } |
| 86 | |
| 87 | armnn::INetworkPtr IOnnxParser::CreateNetworkFromString(const std::string& protoText, |
| 88 | const std::map<std::string, armnn::TensorShape>& inputShapes) |
| 89 | { |
| 90 | return pOnnxParserImpl->CreateNetworkFromString(protoText, inputShapes); |
| 91 | } |
| 92 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 93 | BindingPointInfo IOnnxParser::GetNetworkInputBindingInfo(const std::string& name) const |
| 94 | { |
| 95 | return pOnnxParserImpl->GetNetworkInputBindingInfo(name); |
| 96 | } |
| 97 | |
| 98 | BindingPointInfo IOnnxParser::GetNetworkOutputBindingInfo(const std::string& name) const |
| 99 | { |
| 100 | return pOnnxParserImpl->GetNetworkOutputBindingInfo(name); |
| 101 | } |
| 102 | |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 103 | namespace |
| 104 | { |
| 105 | void CheckValidDataType(std::initializer_list<onnx::TensorProto::DataType> validInputTypes, |
| 106 | const onnx::TensorProto::DataType actualValue, |
| 107 | const char* validExpr, |
| 108 | std::string nodeName, |
| 109 | std::string tensorName, |
| 110 | const armnn::CheckLocation& location) |
| 111 | { |
| 112 | bool isValid = std::any_of(validInputTypes.begin(), |
| 113 | validInputTypes.end(), |
| 114 | [&actualValue](onnx::TensorProto::DataType x) { return x == actualValue; } ); |
| 115 | if (!isValid) |
| 116 | { |
| 117 | throw ParseException( |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 118 | fmt::format("Datatype {} is not valid for tensor '{}' of node '{}', not in {{{}}}. {}", |
| 119 | onnx::TensorProto::DataType_Name(actualValue), |
| 120 | tensorName, |
| 121 | nodeName, |
| 122 | validExpr, |
| 123 | location.AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 124 | } |
| 125 | } |
| 126 | |
| 127 | #define CHECK_VALID_DATATYPE(NODE, TENSOR, ACTUAL, ...) \ |
| 128 | CheckValidDataType({__VA_ARGS__}, ACTUAL, #__VA_ARGS__, NODE, TENSOR, CHECK_LOCATION()) |
| 129 | |
| 130 | using StrTypeListPair = std::pair<const char*, std::initializer_list<onnx::TensorProto::DataType>>; |
| 131 | #define STR_LIST(...) StrTypeListPair(#__VA_ARGS__, {__VA_ARGS__}) |
| 132 | |
| 133 | template <typename Callable> |
| 134 | void ReadMandatoryNodeAttributeImpl(const onnx::NodeProto& node, |
| 135 | const std::string& attribName, |
| 136 | onnx::AttributeProto::AttributeType expectedType, |
| 137 | Callable callable) |
| 138 | { |
| 139 | auto attribs = node.attribute(); |
| 140 | int attriNum = 0; |
| 141 | while (attriNum < node.attribute_size()) |
| 142 | { |
| 143 | if (attribs.Get(attriNum).name() == attribName) |
| 144 | { |
| 145 | if (attribs.Get(attriNum).type() == expectedType) |
| 146 | { |
| 147 | callable(attribs.Get(attriNum)); |
| 148 | } |
| 149 | else |
| 150 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 151 | throw ParseException(fmt::format("Attribute {} of node {} expected to have {} as " |
| 152 | "onnx::AttributeProto::AttributeType, but found {} instead {}", |
| 153 | attribName, |
| 154 | node.name(), |
| 155 | onnx::AttributeProto::AttributeType_Name(expectedType), |
| 156 | onnx::AttributeProto::AttributeType_Name(attribs.Get(attriNum).type()), |
| 157 | CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 158 | } |
| 159 | break; |
| 160 | } |
| 161 | ++attriNum; |
| 162 | } |
| 163 | if (attriNum == node.attribute_size()) |
| 164 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 165 | throw ParseException(fmt::format("Could not find required attribute {} in node {} {}", |
| 166 | attribName, node.name(), CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 167 | } |
| 168 | } |
| 169 | |
| 170 | template <typename Callable> |
| 171 | void ReadOptionalNodeAttributeImpl(const onnx::NodeProto& node, |
| 172 | const std::string& attribName, |
| 173 | onnx::AttributeProto::AttributeType expectedType, |
| 174 | Callable callable) |
| 175 | { |
| 176 | auto attribs = node.attribute(); |
| 177 | for (int attriNum = 0; attriNum < node.attribute_size(); ++attriNum) |
| 178 | { |
| 179 | if (attribs.Get(attriNum).name() == attribName) |
| 180 | { |
| 181 | if (attribs.Get(attriNum).type() == expectedType) |
| 182 | { |
| 183 | callable(attribs.Get(attriNum)); |
| 184 | } |
| 185 | else |
| 186 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 187 | throw ParseException( |
| 188 | fmt::format("Attribute {} of node {} expected to have {} as onnx::AttributeProto::AttributeType, " |
| 189 | "but found {} instead {}", |
| 190 | attribName, |
| 191 | node.name(), |
| 192 | onnx::AttributeProto::AttributeType_Name(expectedType), |
| 193 | onnx::AttributeProto::AttributeType_Name(attribs.Get(attriNum).type()), |
| 194 | CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 195 | } |
| 196 | } |
| 197 | } |
| 198 | } |
| 199 | |
Narumol Prangnawarat | bc3bb62 | 2021-09-24 16:08:34 +0100 | [diff] [blame] | 200 | int ReadMandatoryNodeIntAttribute(const onnx::NodeProto& node, |
| 201 | const std::string& name) |
| 202 | { |
| 203 | int attribValue = 0; |
| 204 | ReadMandatoryNodeAttributeImpl(node, name, onnx::AttributeProto::INT, |
| 205 | [&attribValue](const onnx::AttributeProto& attrValue) |
| 206 | { |
| 207 | attribValue = CHECKED_INT32(attrValue.i()); |
| 208 | }); |
| 209 | return attribValue; |
| 210 | } |
| 211 | |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 212 | int64_t ReadOptionalNodeInt64Attribute(const onnx::NodeProto& node, |
| 213 | const std::string& name, |
| 214 | const int64_t defaultValue = 0) |
| 215 | { |
| 216 | int64_t attribValue = defaultValue; |
| 217 | ReadOptionalNodeAttributeImpl(node, name, onnx::AttributeProto::INT, |
| 218 | [&attribValue](const onnx::AttributeProto& attrValue) |
| 219 | { |
| 220 | attribValue = attrValue.i(); |
| 221 | }); |
| 222 | return attribValue; |
| 223 | } |
| 224 | |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 225 | std::vector<uint32_t> ReadMandatoryNodeUint32ListAttribute(const onnx::NodeProto& node, |
| 226 | const std::string& name) |
| 227 | { |
| 228 | std::vector<uint32_t> attriList; |
| 229 | ReadMandatoryNodeAttributeImpl(node, name, onnx::AttributeProto::INTS, |
| 230 | [&attriList](const onnx::AttributeProto& attrValue) |
| 231 | { |
| 232 | for (int attriNum = 0; attriNum < attrValue.ints_size(); ++attriNum) |
| 233 | { |
| 234 | attriList.push_back(CHECKED_NON_NEGATIVE(CHECKED_INT32(attrValue.ints().Get(attriNum)))); |
| 235 | } |
| 236 | }); |
| 237 | return attriList; |
| 238 | } |
| 239 | |
| 240 | uint32_t ReadOptionalNodeUint32Attribute(const onnx::NodeProto& node, |
| 241 | const std::string& name, |
| 242 | const uint32_t defaultVal = 0u) |
| 243 | { |
| 244 | uint32_t attribValue = defaultVal; |
| 245 | ReadOptionalNodeAttributeImpl(node, name, onnx::AttributeProto::INT, |
| 246 | [&attribValue](const onnx::AttributeProto& attrValue) |
| 247 | { |
| 248 | attribValue = CHECKED_NON_NEGATIVE(CHECKED_INT32((attrValue.i()))); |
| 249 | }); |
| 250 | return attribValue; |
| 251 | } |
| 252 | |
| 253 | std::vector<uint32_t> ReadOptionalNodeUint32ListAttribute(const onnx::NodeProto& node, |
| 254 | const std::string& name) |
| 255 | { |
| 256 | std::vector<uint32_t> attriList; |
| 257 | ReadOptionalNodeAttributeImpl(node, name, onnx::AttributeProto::INTS, |
| 258 | [&attriList](const onnx::AttributeProto& attrValue) |
| 259 | { |
| 260 | for (int attriNum = 0; attriNum < attrValue.ints_size(); ++attriNum) |
| 261 | { |
| 262 | attriList.push_back(CHECKED_NON_NEGATIVE(CHECKED_INT32(attrValue.ints().Get(attriNum)))); |
| 263 | } |
| 264 | }); |
| 265 | |
| 266 | return attriList; |
| 267 | } |
| 268 | |
| 269 | float ReadOptionalNodeFloatAttribute(const onnx::NodeProto& node, |
| 270 | const std::string& name, |
| 271 | const float defaultValue = 0.0f) |
| 272 | { |
| 273 | float attribValue = defaultValue; |
| 274 | ReadOptionalNodeAttributeImpl(node, name, onnx::AttributeProto::FLOAT, |
| 275 | [&attribValue](const onnx::AttributeProto& attrValue) |
| 276 | { |
| 277 | attribValue = attrValue.f(); |
| 278 | }); |
| 279 | return attribValue; |
| 280 | } |
| 281 | |
| 282 | std::string ReadOptionalNodeStringAttribute(const onnx::NodeProto& node, const std::string& name) |
| 283 | { |
| 284 | std::string attribValue = ""; |
| 285 | ReadOptionalNodeAttributeImpl(node, name, onnx::AttributeProto::STRING, |
| 286 | [&attribValue](const onnx::AttributeProto& attrValue) |
| 287 | { |
| 288 | attribValue = attrValue.s(); |
| 289 | }); |
| 290 | return attribValue; |
| 291 | } |
| 292 | |
Tee Jung | fcf6fd5 | 2019-11-01 05:27:28 +0000 | [diff] [blame] | 293 | armnn::TensorInfo ToTensorInfo(const std::string& name, std::vector<unsigned int>& shape, int data_type) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 294 | { |
Narumol Prangnawarat | 452274c | 2021-09-23 16:12:19 +0100 | [diff] [blame] | 295 | DataType type; |
| 296 | switch(data_type) |
| 297 | { |
| 298 | case onnx::TensorProto::FLOAT: |
| 299 | { |
| 300 | type = DataType::Float32; |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 301 | break; |
Narumol Prangnawarat | 452274c | 2021-09-23 16:12:19 +0100 | [diff] [blame] | 302 | } |
| 303 | case onnx::TensorProto::INT32: |
| 304 | case onnx::TensorProto::INT64: |
| 305 | { |
| 306 | type = DataType::Signed32; |
| 307 | break; |
| 308 | } |
| 309 | default: |
| 310 | { |
| 311 | throw ParseException( |
| 312 | fmt::format("'{}' is not a currently supported datatype for tensor {}." |
| 313 | " Supported dataTypes are FLOAT, INT32 and INT64. {}", |
| 314 | onnx::TensorProto::DataType_Name(static_cast<onnx::TensorProto::DataType>(data_type)), |
| 315 | name, |
| 316 | CHECK_LOCATION().AsString() )); |
| 317 | } |
| 318 | } |
Tee Jung | caf2bdd | 2019-11-13 07:23:14 +0000 | [diff] [blame] | 319 | |
Narumol Prangnawarat | 1b11f32 | 2021-10-13 11:44:50 +0100 | [diff] [blame] | 320 | // Scalar Tensor |
Narumol Prangnawarat | 452274c | 2021-09-23 16:12:19 +0100 | [diff] [blame] | 321 | if (shape.empty()) |
| 322 | { |
| 323 | return TensorInfo(TensorShape(Dimensionality::Scalar), type); |
| 324 | } |
Tee Jung | caf2bdd | 2019-11-13 07:23:14 +0000 | [diff] [blame] | 325 | |
Narumol Prangnawarat | 1b11f32 | 2021-10-13 11:44:50 +0100 | [diff] [blame] | 326 | // Dynamic Tensor |
| 327 | if(std::find(shape.begin(), shape.end(), 0) != shape.end()) |
| 328 | { |
| 329 | return TensorInfo(TensorShape(Dimensionality::NotSpecified), type); |
| 330 | } |
| 331 | |
Narumol Prangnawarat | 452274c | 2021-09-23 16:12:19 +0100 | [diff] [blame] | 332 | return TensorInfo(TensorShape(static_cast<unsigned int>(shape.size()), shape.data()), type); |
Tee Jung | fcf6fd5 | 2019-11-01 05:27:28 +0000 | [diff] [blame] | 333 | } |
| 334 | |
| 335 | armnn::TensorInfo ToTensorInfo(const onnx::ValueInfoProto& info) |
| 336 | { |
| 337 | const onnx::TensorShapeProto onnxShape = info.type().tensor_type().shape(); |
| 338 | std::vector<unsigned int> shapeDims; |
| 339 | for (int i = 0; i < onnxShape.dim_size(); ++i) |
| 340 | { |
| 341 | shapeDims.push_back(CHECKED_NON_NEGATIVE(CHECKED_INT32(onnxShape.dim(i).dim_value()))); |
| 342 | } |
| 343 | |
| 344 | return ToTensorInfo(info.name(), shapeDims, info.type().tensor_type().elem_type()); |
| 345 | } |
| 346 | |
| 347 | armnn::TensorInfo ToTensorInfo(const onnx::TensorProto& tensor) |
| 348 | { |
| 349 | std::vector<unsigned int> shapeDims; |
Ryan OShea | 337c17f | 2020-02-21 12:33:17 +0000 | [diff] [blame] | 350 | |
Tee Jung | fcf6fd5 | 2019-11-01 05:27:28 +0000 | [diff] [blame] | 351 | for (auto dim: tensor.dims()) |
| 352 | { |
| 353 | shapeDims.push_back(CHECKED_NON_NEGATIVE(CHECKED_INT32(dim))); |
| 354 | } |
| 355 | |
| 356 | return ToTensorInfo(tensor.name(), shapeDims, tensor.data_type()); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 357 | } |
| 358 | |
| 359 | std::string TensorInfoAsString(const TensorInfo& info, |
| 360 | const std::string& name, |
| 361 | const onnx::TensorProto::DataType& type) |
| 362 | { |
| 363 | const TensorShape shape = info.GetShape(); |
| 364 | std::stringstream ss; |
| 365 | ss << "tensor '" << name << "' contains " |
| 366 | << onnx::TensorProto::DataType_Name(type) |
| 367 | << " and has shape ["; |
| 368 | |
| 369 | for (uint32_t i = 0; i < shape.GetNumDimensions() - 1; ++i) |
| 370 | { |
| 371 | ss << shape[i] << ", "; |
| 372 | } |
| 373 | ss << shape[shape.GetNumDimensions() - 1] << "]"; |
| 374 | return ss.str(); |
| 375 | } |
| 376 | |
Sadik Armagan | 60bb9d8 | 2021-01-11 15:15:01 +0000 | [diff] [blame] | 377 | void CalcPadding(uint32_t inputSize, |
| 378 | uint32_t filterSize, |
| 379 | uint32_t stride, |
| 380 | uint32_t dilation, |
| 381 | uint32_t* paddingFront, |
| 382 | uint32_t* paddingBack, |
| 383 | bool isUpper) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 384 | { |
| 385 | uint32_t outputSize = (inputSize + stride - 1) / stride; |
Sadik Armagan | 60bb9d8 | 2021-01-11 15:15:01 +0000 | [diff] [blame] | 386 | uint32_t dilatedSize = filterSize + (dilation - 1) * (filterSize - 1); |
| 387 | uint32_t temp = (outputSize - 1) * stride + dilatedSize; |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 388 | *paddingFront = (temp - inputSize) / 2; |
| 389 | *paddingBack = *paddingFront; |
| 390 | if((temp - inputSize) % 2 == 1) |
| 391 | { |
| 392 | if (isUpper) |
| 393 | { |
Sadik Armagan | 60bb9d8 | 2021-01-11 15:15:01 +0000 | [diff] [blame] | 394 | *paddingBack += 1; |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 395 | } |
| 396 | else |
| 397 | { |
Sadik Armagan | 60bb9d8 | 2021-01-11 15:15:01 +0000 | [diff] [blame] | 398 | *paddingFront += 1; |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 399 | } |
| 400 | } |
| 401 | } |
| 402 | |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 403 | TensorInfo ComputeReshapeInfo(const TensorShape& targetShapeTensor, |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 404 | const TensorShape& inShape, |
Narumol Prangnawarat | 452274c | 2021-09-23 16:12:19 +0100 | [diff] [blame] | 405 | const std::string& outName, |
| 406 | DataType dataType = DataType::Float32) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 407 | { |
| 408 | std::vector<int> targetDims; |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 409 | for(uint i = 0; i < targetShapeTensor.GetNumDimensions(); ++i) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 410 | { |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 411 | int val = CHECKED_INT32(targetShapeTensor[i]); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 412 | if(val == 0) |
| 413 | { |
| 414 | targetDims.push_back(static_cast<int>(inShape[static_cast<uint>(i)])); |
| 415 | } |
| 416 | else |
| 417 | { |
| 418 | targetDims.push_back(val); |
| 419 | } |
| 420 | } |
| 421 | |
| 422 | std::vector<unsigned int> outDims(targetDims.begin(), targetDims.end()); |
| 423 | const auto stretchDim = std::find(targetDims.begin(), targetDims.end(), -1); |
| 424 | if (stretchDim != targetDims.end()) |
| 425 | { |
| 426 | if (std::find(std::next(stretchDim), targetDims.end(), -1) != targetDims.end()) |
| 427 | { |
| 428 | std::stringstream ss; |
| 429 | ss << "[ "; |
| 430 | for(uint i = 0; i < targetDims.size() - 1; ++i) |
| 431 | { |
| 432 | ss << targetDims[i] << ", "; |
| 433 | } |
| 434 | ss << targetDims[targetDims.size() - 1] << " ]"; |
| 435 | |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 436 | throw ParseException( |
| 437 | fmt::format("Error during creation of reshaped tensor '{}'. At most one component of shape can be " |
| 438 | " -1 and here, shape is {} {}", |
| 439 | outName, |
| 440 | ss.str(), |
| 441 | CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 442 | } |
| 443 | |
Matthew Sloyan | 589e3e8 | 2020-09-11 16:17:48 +0100 | [diff] [blame] | 444 | auto targetNumElements = armnn::numeric_cast<unsigned int>(std::accumulate(targetDims.begin(), targetDims.end(), |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 445 | -1, std::multiplies<int32_t>())); |
| 446 | auto stretchIndex = static_cast<size_t>(std::distance(targetDims.begin(), stretchDim)); |
| 447 | outDims[stretchIndex] = inShape.GetNumElements() / targetNumElements; |
| 448 | } |
| 449 | TensorShape outShape = TensorShape{static_cast<unsigned int>(outDims.size()), outDims.data()}; |
Narumol Prangnawarat | 452274c | 2021-09-23 16:12:19 +0100 | [diff] [blame] | 450 | return TensorInfo(outShape, dataType); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 451 | } |
| 452 | |
| 453 | } //namespace |
| 454 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 455 | const std::map<std::string, OnnxParserImpl::OperationParsingFunction> OnnxParserImpl::m_ParserFunctions = { |
| 456 | { "BatchNormalization", &OnnxParserImpl::ParseBatchNormalization}, |
| 457 | { "GlobalAveragePool", &OnnxParserImpl::ParseGlobalAveragePool}, |
| 458 | { "AveragePool", &OnnxParserImpl::ParseAveragePool }, |
| 459 | { "Clip", &OnnxParserImpl::ParseClip }, |
| 460 | { "Constant", &OnnxParserImpl::ParseConstant }, |
| 461 | { "MaxPool", &OnnxParserImpl::ParseMaxPool }, |
| 462 | { "Reshape", &OnnxParserImpl::ParseReshape }, |
| 463 | { "Sigmoid", &OnnxParserImpl::ParseSigmoid }, |
| 464 | { "Tanh", &OnnxParserImpl::ParseTanh }, |
| 465 | { "Relu", &OnnxParserImpl::ParseRelu }, |
| 466 | { "LeakyRelu", &OnnxParserImpl::ParseLeakyRelu }, |
| 467 | { "Conv", &OnnxParserImpl::ParseConv }, |
| 468 | { "Add", &OnnxParserImpl::ParseAdd }, |
Narumol Prangnawarat | cdc495e | 2021-09-16 18:13:39 +0100 | [diff] [blame] | 469 | { "Flatten", &OnnxParserImpl::ParseFlatten }, |
Narumol Prangnawarat | f10b15a | 2021-09-17 21:08:57 +0100 | [diff] [blame] | 470 | { "Shape", &OnnxParserImpl::ParseShape }, |
| 471 | { "Gather", &OnnxParserImpl::ParseGather }, |
Narumol Prangnawarat | bc3bb62 | 2021-09-24 16:08:34 +0100 | [diff] [blame] | 472 | { "Unsqueeze", &OnnxParserImpl::ParseUnsqueeze }, |
Narumol Prangnawarat | 1112b01 | 2021-09-30 12:10:50 +0100 | [diff] [blame] | 473 | { "Concat", &OnnxParserImpl::ParseConcat }, |
| 474 | { "Gemm", &OnnxParserImpl::ParseGemm } |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 475 | }; |
| 476 | |
| 477 | template<typename TypePair, typename Location> |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 478 | void OnnxParserImpl::ValidateInputs(const onnx::NodeProto& node, |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 479 | TypePair validInputs, |
| 480 | const Location& location) |
| 481 | { |
| 482 | for(auto input : node.input()) |
| 483 | { |
| 484 | CheckValidDataType(validInputs.second, |
| 485 | m_TensorsInfo[input].m_dtype, |
| 486 | validInputs.first, |
| 487 | node.name(), |
| 488 | input, |
| 489 | location); |
| 490 | } |
| 491 | } |
| 492 | |
| 493 | #define VALID_INPUTS(NODE, VALID_INPUTS) \ |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 494 | OnnxParserImpl::ValidateInputs(NODE, \ |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 495 | VALID_INPUTS, \ |
| 496 | CHECK_LOCATION()) |
| 497 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 498 | std::vector<TensorInfo> OnnxParserImpl::ComputeOutputInfo(std::vector<std::string> outNames, |
| 499 | const IConnectableLayer* layer, |
Narumol Prangnawarat | 452274c | 2021-09-23 16:12:19 +0100 | [diff] [blame] | 500 | std::vector<TensorShape> inputShapes, |
| 501 | const onnx::TensorProto::DataType& dataType) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 502 | { |
Narumol Prangnawarat | ac2770a | 2020-04-01 16:51:23 +0100 | [diff] [blame] | 503 | ARMNN_ASSERT(! outNames.empty()); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 504 | bool needCompute = std::any_of(outNames.begin(), |
| 505 | outNames.end(), |
| 506 | [this](std::string name) |
| 507 | { |
Matthew Sloyan | ca36123 | 2023-02-16 14:50:22 +0000 | [diff] [blame] | 508 | return (m_TensorsInfo.count(name) == 0 || |
| 509 | m_TensorsInfo[name].m_info == nullptr || |
| 510 | m_TensorsInfo[name].m_info->GetShape().GetDimensionality() == |
| 511 | Dimensionality::NotSpecified); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 512 | }); |
Narumol Prangnawarat | 452274c | 2021-09-23 16:12:19 +0100 | [diff] [blame] | 513 | std::vector<TensorInfo> outInfo; |
| 514 | //if the output info(s) are not here, we need to compute them |
| 515 | std::vector<TensorShape> inferredShapes; |
| 516 | DataType armnnType = DataType::Float32; |
| 517 | if(needCompute) { |
| 518 | inferredShapes = layer->InferOutputShapes(inputShapes); |
| 519 | ARMNN_ASSERT(inferredShapes.size() == outNames.size()); |
| 520 | switch (dataType) { |
| 521 | case onnx::TensorProto::FLOAT: { |
| 522 | armnnType = DataType::Float32; |
| 523 | break; |
| 524 | } |
| 525 | case onnx::TensorProto::INT32: |
| 526 | case onnx::TensorProto::INT64: { |
| 527 | armnnType = DataType::Signed32; |
| 528 | break; |
| 529 | } |
| 530 | default: { |
| 531 | throw ParseException( |
| 532 | fmt::format("'{}' is not a currently supported datatype for {}." |
| 533 | " Supported dataTypes are FLOAT, INT32 and INT64. {}", |
| 534 | onnx::TensorProto::DataType_Name(static_cast<onnx::TensorProto::DataType>(dataType)), |
| 535 | layer->GetName(), |
| 536 | CHECK_LOCATION().AsString())); |
| 537 | } |
| 538 | } |
| 539 | } |
| 540 | for (uint i = 0; i < outNames.size(); ++i) |
| 541 | { |
| 542 | if(needCompute) |
| 543 | { |
| 544 | m_TensorsInfo[outNames[i]] = OnnxTensor(); |
| 545 | m_TensorsInfo[outNames[i]].m_info = std::make_unique<TensorInfo>( |
| 546 | TensorInfo(inferredShapes[i], armnnType)); |
| 547 | m_TensorsInfo[outNames[i]].m_dtype = dataType; |
| 548 | } |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 549 | outInfo.push_back(*m_TensorsInfo[outNames[i]].m_info); |
Narumol Prangnawarat | 452274c | 2021-09-23 16:12:19 +0100 | [diff] [blame] | 550 | } |
| 551 | return outInfo; |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 552 | } |
| 553 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 554 | OnnxParserImpl::OnnxParserImpl() |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 555 | : m_Network(nullptr, nullptr) |
| 556 | { |
| 557 | } |
| 558 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 559 | void OnnxParserImpl::ResetParser() |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 560 | { |
| 561 | m_Network = armnn::INetworkPtr(nullptr, nullptr); |
| 562 | m_Graph = nullptr; |
Narumol Prangnawarat | 1b11f32 | 2021-10-13 11:44:50 +0100 | [diff] [blame] | 563 | m_InputInfos.clear(); |
| 564 | m_OutputInfos.clear(); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 565 | } |
| 566 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 567 | void OnnxParserImpl::Cleanup() |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 568 | { |
| 569 | m_TensorConnections.clear(); |
| 570 | m_TensorsInfo.clear(); |
| 571 | m_OutputsMap.clear(); |
| 572 | m_OutputsFusedAndUsed.clear(); |
Narumol Prangnawarat | 1b11f32 | 2021-10-13 11:44:50 +0100 | [diff] [blame] | 573 | m_InputShapes.clear(); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 574 | } |
| 575 | |
Jan Eilers | 53ef795 | 2021-06-02 12:01:25 +0100 | [diff] [blame] | 576 | template<typename T> |
| 577 | std::pair<armnn::ConstTensor, std::unique_ptr<T[]>> |
| 578 | CreateConstTensorImpl(const T* bufferPtr, |
| 579 | armnn::TensorInfo& tensorInfo, |
| 580 | const armnn::Optional<armnn::PermutationVector&> permutationVector) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 581 | { |
Jan Eilers | 53ef795 | 2021-06-02 12:01:25 +0100 | [diff] [blame] | 582 | ARMNN_ASSERT_MSG(bufferPtr != nullptr, fmt::format("Buffer for permutation is null").c_str()); |
| 583 | |
| 584 | std::unique_ptr<T[]> data(new T[tensorInfo.GetNumElements()]); |
| 585 | |
| 586 | if (permutationVector.has_value() && permutationVector.value().GetSize() > 0) |
| 587 | { |
| 588 | tensorInfo = armnnUtils::Permuted(tensorInfo, permutationVector.value()); |
| 589 | armnnUtils::Permute(tensorInfo.GetShape(), permutationVector.value(), |
| 590 | reinterpret_cast<const T*>(bufferPtr), data.get(), sizeof(T)); |
| 591 | } |
| 592 | else |
| 593 | { |
| 594 | ::memcpy(data.get(), bufferPtr, tensorInfo.GetNumBytes()); |
| 595 | } |
| 596 | |
| 597 | return std::make_pair(ConstTensor(tensorInfo, data.get()), std::move(data)); |
| 598 | } |
| 599 | |
| 600 | std::pair<ConstTensor, std::unique_ptr<float[]>> |
| 601 | OnnxParserImpl::CreateConstTensor(const std::string name, |
| 602 | armnn::Optional<armnn::PermutationVector&> permutationVector) |
| 603 | { |
| 604 | TensorInfo tensorInfo = *m_TensorsInfo[name].m_info; |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 605 | onnx::TensorProto onnxTensor = *m_TensorsInfo[name].m_tensor; |
| 606 | |
Narumol Prangnawarat | f10b15a | 2021-09-17 21:08:57 +0100 | [diff] [blame] | 607 | //ONNX can have Float16 and double constant nodes but ArmNN only supports float32 |
| 608 | CHECK_VALID_DATATYPE(name, onnxTensor.name(), |
| 609 | static_cast<onnx::TensorProto::DataType>(onnxTensor.data_type()), onnx::TensorProto::FLOAT); |
| 610 | |
Matthew Sloyan | 81beae3 | 2021-07-13 19:46:11 +0100 | [diff] [blame] | 611 | // Makes sure IsConstant flag is set. |
| 612 | tensorInfo.SetConstant(); |
| 613 | |
Jan Eilers | 53ef795 | 2021-06-02 12:01:25 +0100 | [diff] [blame] | 614 | // Const tensors requires at least a list of values |
| 615 | if (tensorInfo.GetNumElements() == 0) |
| 616 | { |
| 617 | throw ParseException(fmt::format("No tensor data found for Const tensor '{}' {}", |
| 618 | name, |
| 619 | CHECK_LOCATION().AsString())); |
| 620 | } |
| 621 | |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 622 | auto srcData = onnxTensor.float_data().data(); |
Pablo Tello | 3dcc1c6 | 2019-04-24 14:20:21 +0100 | [diff] [blame] | 623 | // Copy the value list entries into the destination |
| 624 | if (!onnxTensor.has_raw_data()) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 625 | { |
Pablo Tello | 3dcc1c6 | 2019-04-24 14:20:21 +0100 | [diff] [blame] | 626 | if(tensorInfo.GetNumElements() != static_cast<uint>(onnxTensor.float_data_size())) |
| 627 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 628 | throw ParseException( |
| 629 | fmt::format("The number of data provided ({}) does not match the tensor '{}' number of " |
| 630 | "elements ({}) {}", |
| 631 | onnxTensor.float_data_size(), |
| 632 | name, |
| 633 | tensorInfo.GetNumElements(), |
| 634 | CHECK_LOCATION().AsString())); |
Pablo Tello | 3dcc1c6 | 2019-04-24 14:20:21 +0100 | [diff] [blame] | 635 | } |
Jan Eilers | 53ef795 | 2021-06-02 12:01:25 +0100 | [diff] [blame] | 636 | return CreateConstTensorImpl<float>(srcData, tensorInfo, permutationVector); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 637 | } |
Pablo Tello | 3dcc1c6 | 2019-04-24 14:20:21 +0100 | [diff] [blame] | 638 | else |
| 639 | { |
Jan Eilers | 53ef795 | 2021-06-02 12:01:25 +0100 | [diff] [blame] | 640 | return CreateConstTensorImpl<float>(reinterpret_cast<const float*>(onnxTensor.raw_data().c_str()), |
| 641 | tensorInfo, |
| 642 | permutationVector); |
Pablo Tello | 3dcc1c6 | 2019-04-24 14:20:21 +0100 | [diff] [blame] | 643 | } |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 644 | } |
| 645 | |
Narumol Prangnawarat | f10b15a | 2021-09-17 21:08:57 +0100 | [diff] [blame] | 646 | std::pair<ConstTensor, std::unique_ptr<int32_t[]>> |
| 647 | OnnxParserImpl::CreateInt64ConstTensor(const std::string name, |
| 648 | armnn::Optional<armnn::PermutationVector&> permutationVector) |
| 649 | { |
| 650 | TensorInfo tensorInfo = *m_TensorsInfo[name].m_info; |
| 651 | onnx::TensorProto onnxTensor = *m_TensorsInfo[name].m_tensor; |
| 652 | |
| 653 | CHECK_VALID_DATATYPE(name, onnxTensor.name(), |
| 654 | static_cast<onnx::TensorProto::DataType>(onnxTensor.data_type()), onnx::TensorProto::INT64); |
| 655 | |
| 656 | // Makes sure IsConstant flag is set. |
| 657 | tensorInfo.SetConstant(); |
| 658 | uint numElements = tensorInfo.GetNumElements(); |
| 659 | |
| 660 | // Const tensors requires at least a list of values |
| 661 | if (numElements == 0) |
| 662 | { |
| 663 | throw ParseException(fmt::format("No tensor data found for Const tensor '{}' {}", |
| 664 | name, |
| 665 | CHECK_LOCATION().AsString())); |
| 666 | } |
| 667 | |
| 668 | // Copy the value list entries into the destination |
| 669 | if (!onnxTensor.has_raw_data()) |
| 670 | { |
| 671 | auto srcData = onnxTensor.int64_data().data(); |
| 672 | if(numElements != static_cast<uint>(onnxTensor.int64_data_size())) |
| 673 | { |
| 674 | throw ParseException( |
| 675 | fmt::format("The number of data provided ({}) does not match the tensor '{}' number of " |
| 676 | "elements ({}) {}", |
| 677 | onnxTensor.int64_data_size(), |
| 678 | name, |
| 679 | tensorInfo.GetNumElements(), |
| 680 | CHECK_LOCATION().AsString())); |
| 681 | } |
| 682 | |
| 683 | std::vector<int32_t> int32Data; |
| 684 | for(uint i = 0; i < numElements; i++) |
| 685 | { |
| 686 | int32_t int32Value = CHECKED_INT32(srcData[i]); |
| 687 | int32Data.push_back(int32Value); |
| 688 | } |
| 689 | |
| 690 | return CreateConstTensorImpl<int32_t>(int32Data.data(), tensorInfo, permutationVector); |
| 691 | } |
| 692 | else |
| 693 | { |
| 694 | auto srcData = reinterpret_cast<const int64_t*>(onnxTensor.raw_data().c_str()); |
| 695 | std::vector<int32_t> int32Data; |
| 696 | for(uint i = 0; i < numElements; i++) |
| 697 | { |
| 698 | int32_t int32Value = CHECKED_INT32(srcData[i]); |
| 699 | int32Data.push_back(int32Value); |
| 700 | } |
| 701 | return CreateConstTensorImpl<int32_t>(int32Data.data(), tensorInfo, permutationVector); |
| 702 | } |
| 703 | } |
| 704 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 705 | ModelPtr OnnxParserImpl::LoadModelFromTextFile(const char* graphFile) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 706 | { |
| 707 | FILE* fd = fopen(graphFile, "r"); |
| 708 | |
| 709 | if (fd == nullptr) |
| 710 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 711 | throw FileNotFoundException(fmt::format("Invalid (null) filename {}", CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 712 | } |
| 713 | |
| 714 | // Parse the file into a message |
| 715 | ModelPtr modelProto = std::make_unique<onnx::ModelProto>(); |
| 716 | using google::protobuf::io::FileInputStream; |
| 717 | std::unique_ptr<FileInputStream> input = std::make_unique<FileInputStream>(fileno(fd)); |
| 718 | bool success = google::protobuf::TextFormat::Parse(input.get(), modelProto.get()); |
| 719 | fclose(fd); |
| 720 | |
| 721 | if (!success) |
| 722 | { |
| 723 | std::stringstream error; |
| 724 | error << "Failed to parse graph file"; |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 725 | throw ParseException(fmt::format("{} {}", error.str(), CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 726 | } |
| 727 | return modelProto; |
| 728 | } |
| 729 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 730 | INetworkPtr OnnxParserImpl::CreateNetworkFromTextFile(const char* graphFile) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 731 | { |
| 732 | ResetParser(); |
| 733 | ModelPtr modelProto = LoadModelFromTextFile(graphFile); |
| 734 | return CreateNetworkFromModel(*modelProto); |
| 735 | } |
| 736 | |
Narumol Prangnawarat | 1b11f32 | 2021-10-13 11:44:50 +0100 | [diff] [blame] | 737 | INetworkPtr OnnxParserImpl::CreateNetworkFromTextFile(const char* graphFile, |
| 738 | const std::map<std::string, armnn::TensorShape>& inputShapes) |
| 739 | { |
| 740 | ResetParser(); |
| 741 | m_InputShapes = inputShapes; |
| 742 | ModelPtr modelProto = LoadModelFromTextFile(graphFile); |
| 743 | return CreateNetworkFromModel(*modelProto); |
| 744 | } |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 745 | |
Mike Kelly | 2ae3224 | 2022-11-25 13:55:24 +0000 | [diff] [blame] | 746 | INetworkPtr OnnxParserImpl::CreateNetworkFromBinary(const std::vector<uint8_t>& binaryContent) |
| 747 | { |
| 748 | ResetParser(); |
| 749 | ModelPtr modelProto = LoadModelFromBinary(binaryContent); |
| 750 | return CreateNetworkFromModel(*modelProto); |
| 751 | } |
| 752 | |
| 753 | INetworkPtr OnnxParserImpl::CreateNetworkFromBinary(const std::vector<uint8_t>& binaryContent, |
| 754 | const std::map<std::string, armnn::TensorShape>& inputShapes) |
| 755 | { |
| 756 | ResetParser(); |
| 757 | m_InputShapes = inputShapes; |
| 758 | ModelPtr modelProto = LoadModelFromBinary(binaryContent); |
| 759 | return CreateNetworkFromModel(*modelProto); |
| 760 | } |
| 761 | |
| 762 | ModelPtr OnnxParserImpl::LoadModelFromBinary(const std::vector<uint8_t>& binaryContent) |
| 763 | { |
| 764 | if (binaryContent.size() == 0) |
| 765 | { |
| 766 | throw ParseException(fmt::format("Missing binary content", CHECK_LOCATION().AsString())); |
| 767 | } |
| 768 | // Parse the file into a message |
| 769 | ModelPtr modelProto = std::make_unique<onnx::ModelProto>(); |
| 770 | |
| 771 | google::protobuf::io::CodedInputStream codedStream(binaryContent.data(), static_cast<int>(binaryContent.size())); |
| 772 | codedStream.SetTotalBytesLimit(INT_MAX); |
| 773 | bool success = modelProto.get()->ParseFromCodedStream(&codedStream); |
| 774 | |
| 775 | if (!success) |
| 776 | { |
| 777 | std::stringstream error; |
| 778 | error << "Failed to parse graph"; |
| 779 | throw ParseException(fmt::format("{} {}", error.str(), CHECK_LOCATION().AsString())); |
| 780 | } |
| 781 | return modelProto; |
| 782 | } |
| 783 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 784 | ModelPtr OnnxParserImpl::LoadModelFromBinaryFile(const char* graphFile) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 785 | { |
| 786 | FILE* fd = fopen(graphFile, "rb"); |
| 787 | |
| 788 | if (fd == nullptr) |
| 789 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 790 | throw FileNotFoundException(fmt::format("Invalid (null) filename {}", CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 791 | } |
| 792 | |
| 793 | // Parse the file into a message |
| 794 | ModelPtr modelProto = std::make_unique<onnx::ModelProto>(); |
| 795 | |
| 796 | google::protobuf::io::FileInputStream inStream(fileno(fd)); |
| 797 | google::protobuf::io::CodedInputStream codedStream(&inStream); |
Nikhil Raj | e518153 | 2020-10-09 14:52:25 +0100 | [diff] [blame] | 798 | codedStream.SetTotalBytesLimit(INT_MAX); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 799 | bool success = modelProto.get()->ParseFromCodedStream(&codedStream); |
| 800 | fclose(fd); |
| 801 | |
| 802 | if (!success) |
| 803 | { |
| 804 | std::stringstream error; |
| 805 | error << "Failed to parse graph file"; |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 806 | throw ParseException(fmt::format("{} {}", error.str(), CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 807 | } |
| 808 | return modelProto; |
| 809 | |
| 810 | } |
| 811 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 812 | INetworkPtr OnnxParserImpl::CreateNetworkFromBinaryFile(const char* graphFile) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 813 | { |
| 814 | ResetParser(); |
| 815 | ModelPtr modelProto = LoadModelFromBinaryFile(graphFile); |
| 816 | return CreateNetworkFromModel(*modelProto); |
| 817 | } |
| 818 | |
Narumol Prangnawarat | 1b11f32 | 2021-10-13 11:44:50 +0100 | [diff] [blame] | 819 | INetworkPtr OnnxParserImpl::CreateNetworkFromBinaryFile(const char* graphFile, |
| 820 | const std::map<std::string, armnn::TensorShape>& inputShapes) |
| 821 | { |
| 822 | ResetParser(); |
| 823 | m_InputShapes = inputShapes; |
| 824 | ModelPtr modelProto = LoadModelFromBinaryFile(graphFile); |
| 825 | return CreateNetworkFromModel(*modelProto); |
| 826 | } |
| 827 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 828 | ModelPtr OnnxParserImpl::LoadModelFromString(const std::string& protoText) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 829 | { |
| 830 | if (protoText == "") |
| 831 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 832 | throw InvalidArgumentException(fmt::format("Invalid (empty) string for model parameter {}", |
| 833 | CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 834 | } |
| 835 | // Parse the string into a message |
| 836 | ModelPtr modelProto = std::make_unique<onnx::ModelProto>(); |
| 837 | bool success = google::protobuf::TextFormat::ParseFromString(protoText, modelProto.get()); |
| 838 | if (!success) |
| 839 | { |
| 840 | std::stringstream error; |
| 841 | error << "Failed to parse graph file"; |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 842 | throw ParseException(fmt::format("{} {}", error.str(), CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 843 | } |
| 844 | return modelProto; |
| 845 | } |
| 846 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 847 | INetworkPtr OnnxParserImpl::CreateNetworkFromString(const std::string& protoText) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 848 | { |
| 849 | ResetParser(); |
| 850 | ModelPtr modelProto = LoadModelFromString(protoText); |
| 851 | return CreateNetworkFromModel(*modelProto); |
| 852 | } |
| 853 | |
Narumol Prangnawarat | 1b11f32 | 2021-10-13 11:44:50 +0100 | [diff] [blame] | 854 | INetworkPtr OnnxParserImpl::CreateNetworkFromString(const std::string& protoText, |
| 855 | const std::map<std::string, armnn::TensorShape>& inputShapes) |
| 856 | { |
| 857 | ResetParser(); |
| 858 | m_InputShapes = inputShapes; |
| 859 | ModelPtr modelProto = LoadModelFromString(protoText); |
| 860 | return CreateNetworkFromModel(*modelProto); |
| 861 | } |
| 862 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 863 | INetworkPtr OnnxParserImpl::CreateNetworkFromModel(onnx::ModelProto& model) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 864 | { |
| 865 | m_Network = INetwork::Create(); |
| 866 | try |
| 867 | { |
| 868 | m_Graph = std::make_unique<onnx::GraphProto>(*model.mutable_graph()); |
| 869 | LoadGraph(); |
| 870 | } |
| 871 | catch (const ParseException& e) |
| 872 | { |
| 873 | Cleanup(); |
| 874 | throw e; |
| 875 | } |
| 876 | Cleanup(); |
| 877 | return std::move(m_Network); |
| 878 | } |
| 879 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 880 | void OnnxParserImpl::LoadGraph() |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 881 | { |
Narumol Prangnawarat | ac2770a | 2020-04-01 16:51:23 +0100 | [diff] [blame] | 882 | ARMNN_ASSERT(m_Graph.get() != nullptr); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 883 | |
| 884 | //Fill m_TensorsInfo with the shapes and value of every tensor |
| 885 | SetupInfo(m_Graph->mutable_output()); |
| 886 | SetupInfo(m_Graph->mutable_input()); |
| 887 | SetupInfo(m_Graph->mutable_value_info()); |
| 888 | |
| 889 | for (auto tensor : m_Graph->initializer()) |
| 890 | { |
| 891 | m_TensorsInfo[tensor.name()].m_tensor = std::make_unique<const onnx::TensorProto>(tensor); |
Tee Jung | fcf6fd5 | 2019-11-01 05:27:28 +0000 | [diff] [blame] | 892 | m_TensorsInfo[tensor.name()].m_info = std::make_unique<TensorInfo>(ToTensorInfo(tensor)); |
| 893 | m_TensorsInfo[tensor.name()].m_dtype = |
| 894 | static_cast<onnx::TensorProto::DataType>(tensor.data_type()); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 895 | } |
| 896 | |
| 897 | SetupInputLayers(); |
| 898 | SetupOutputLayers(); |
| 899 | |
| 900 | //Detect FullyConnected layers with bias and update the FusedAndUsed map acccordingly |
| 901 | DetectFullyConnected(); |
| 902 | |
| 903 | //Parsing the graph |
| 904 | for(size_t nodeIndex = 0; nodeIndex < static_cast<size_t>(m_Graph->node_size()); nodeIndex++) |
| 905 | { |
| 906 | auto node = m_Graph->node(static_cast<int>(nodeIndex)); |
| 907 | const std::string& operation = node.op_type(); |
| 908 | |
| 909 | // check which layers we handled already (add and matmul fused as FC) |
Ryan OShea | 337c17f | 2020-02-21 12:33:17 +0000 | [diff] [blame] | 910 | if (operation == "MatMul" ) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 911 | { |
| 912 | if(m_OutputsFusedAndUsed[nodeIndex].inputForNodes != m_OutputsFusedAndUsed[nodeIndex].fusedWithNodes.size()) |
| 913 | { |
| 914 | //Node which can not be fused as a FullyConnected layer (used in layers as a simple matmul output) |
| 915 | AddFullyConnected(node); |
| 916 | } |
| 917 | } |
| 918 | else if (!(m_OutputsFusedAndUsed[nodeIndex].fusedWithNodes.empty()) && operation == "Add") |
| 919 | { |
| 920 | int matmulIndex = static_cast<int> (m_OutputsFusedAndUsed[nodeIndex].fusedWithNodes[0]); |
| 921 | AddFullyConnected(m_Graph->node(matmulIndex), &node); |
| 922 | } |
| 923 | else if (m_OutputsFusedAndUsed[nodeIndex].fusedWithNodes.empty()) //node is not part of a fused layer |
| 924 | { |
| 925 | auto it = m_ParserFunctions.find(operation); |
| 926 | if (it != m_ParserFunctions.end()) |
| 927 | { |
| 928 | auto func = it->second; |
| 929 | (this->*func)(node); |
| 930 | } |
| 931 | else |
| 932 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 933 | throw ParseException(fmt::format("Unsupported operation {} for node '{}' {}", |
| 934 | operation, |
| 935 | node.name(), |
| 936 | CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 937 | } |
| 938 | } |
| 939 | } |
| 940 | |
| 941 | //Making the connections between outputs and inputs of each layers |
| 942 | for (const auto& tensorCon : m_TensorConnections) |
| 943 | { |
| 944 | if (tensorCon.second.outputSlot != nullptr) |
| 945 | { |
| 946 | for (size_t inputSlotIdx = 0; inputSlotIdx < tensorCon.second.inputSlots.size(); ++inputSlotIdx) |
| 947 | { |
| 948 | tensorCon.second.outputSlot->Connect(*(tensorCon.second.inputSlots[inputSlotIdx])); |
| 949 | } |
| 950 | } |
| 951 | } |
Narumol Prangnawarat | 1b11f32 | 2021-10-13 11:44:50 +0100 | [diff] [blame] | 952 | |
| 953 | // Get output info. |
| 954 | for(int outputIndex = 0; outputIndex < m_Graph->output_size(); ++outputIndex) |
| 955 | { |
| 956 | auto output = m_Graph->output(outputIndex); |
| 957 | m_OutputInfos[output.name()] = *m_TensorsInfo[output.name()].m_info; |
| 958 | } |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 959 | } |
| 960 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 961 | void OnnxParserImpl::SetupInfo(const google::protobuf::RepeatedPtrField<onnx::ValueInfoProto >* list) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 962 | { |
| 963 | for (auto tensor : *list) |
| 964 | { |
| 965 | m_TensorsInfo[tensor.name()] = OnnxTensor(); |
| 966 | m_TensorsInfo[tensor.name()].m_info = std::make_unique<TensorInfo>(ToTensorInfo(tensor)); |
Matteo Martincigh | e355dc2 | 2018-12-10 13:45:27 +0000 | [diff] [blame] | 967 | m_TensorsInfo[tensor.name()].m_dtype = |
| 968 | static_cast<onnx::TensorProto::DataType>(tensor.type().tensor_type().elem_type()); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 969 | } |
| 970 | } |
| 971 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 972 | void OnnxParserImpl::DetectFullyConnected() |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 973 | { |
| 974 | m_OutputsFusedAndUsed = std::vector<UsageSummary> (static_cast<size_t>(m_Graph->node_size()), UsageSummary()); |
| 975 | auto matmulAndConstant = [&](const std::string& constInput, |
| 976 | const std::string& matmulInput, |
| 977 | int& nodeIndex) |
| 978 | { |
| 979 | auto matmulIt = m_OutputsMap.find(matmulInput); |
| 980 | if(matmulIt != m_OutputsMap.end() && matmulIt->second.first->op_type() == "MatMul" |
| 981 | && m_TensorsInfo[constInput].isConstant()) |
| 982 | { |
| 983 | nodeIndex = matmulIt->second.second; |
| 984 | return true; |
| 985 | } |
| 986 | return false; |
| 987 | }; |
| 988 | |
| 989 | for(int nodeIndex = 0; nodeIndex < m_Graph->node_size(); nodeIndex++) |
| 990 | { |
| 991 | const onnx::NodeProto* node = &m_Graph->node(nodeIndex); |
| 992 | for (const std::string& output : node->output()) |
| 993 | { |
| 994 | m_OutputsMap[output] = std::make_pair(node, nodeIndex); |
| 995 | } |
| 996 | |
| 997 | for (const std::string& input : node->input()) //count how many time a node is used as input |
| 998 | { |
| 999 | auto matmulIt = m_OutputsMap.find(input); |
| 1000 | if(matmulIt != m_OutputsMap.end()){ |
| 1001 | ++m_OutputsFusedAndUsed[static_cast<size_t>(matmulIt->second.second)].inputForNodes; //node used |
| 1002 | } |
| 1003 | } |
| 1004 | |
| 1005 | if (node->op_type() == "Add") |
| 1006 | { |
| 1007 | int matmulIndex = 0; |
| 1008 | if (matmulAndConstant(node->input(0), node->input(1), matmulIndex) || |
| 1009 | matmulAndConstant(node->input(1), node->input(0), matmulIndex)) |
| 1010 | { |
| 1011 | //matmul and add were fused |
| 1012 | m_OutputsFusedAndUsed[static_cast<size_t>(matmulIndex)].fusedWithNodes |
| 1013 | .push_back(static_cast<size_t>(nodeIndex)); |
| 1014 | |
| 1015 | m_OutputsFusedAndUsed[static_cast<size_t>(nodeIndex)].fusedWithNodes |
| 1016 | .push_back(static_cast<size_t>(matmulIndex)); |
| 1017 | } |
| 1018 | } |
| 1019 | } |
| 1020 | |
| 1021 | for (auto output: m_Graph->output()) { //Add usages as output of the graph in count of usages |
| 1022 | auto matmulIt = m_OutputsMap.find(output.name()); |
| 1023 | if(matmulIt != m_OutputsMap.end()){ |
| 1024 | ++m_OutputsFusedAndUsed[static_cast<size_t>(matmulIt->second.second)].inputForNodes; |
| 1025 | } |
| 1026 | } |
| 1027 | } |
| 1028 | |
| 1029 | template<typename Location> |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 1030 | void OnnxParserImpl::GetInputAndParam(const onnx::NodeProto& node, |
| 1031 | std::string* inputName, |
| 1032 | std::string* constName, |
| 1033 | const Location& location) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1034 | { |
| 1035 | int cstIndex; |
| 1036 | if (m_TensorsInfo[node.input(0)].isConstant()) |
| 1037 | { |
| 1038 | cstIndex = 0; |
| 1039 | } |
| 1040 | else if (m_TensorsInfo[node.input(1)].isConstant()) |
| 1041 | { |
| 1042 | cstIndex = 1; |
| 1043 | } |
| 1044 | else |
| 1045 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 1046 | throw ParseException(fmt::format("One of the input tensors ('{}' or '{}') should be constant in node '{}' {}", |
| 1047 | node.input(0), |
| 1048 | node.input(1), |
| 1049 | node.name(), |
| 1050 | location.AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1051 | } |
| 1052 | if(constName) |
| 1053 | { |
| 1054 | *constName = node.input(cstIndex); |
| 1055 | } |
| 1056 | if(inputName) |
| 1057 | { |
| 1058 | *inputName = node.input(!cstIndex); |
| 1059 | } |
| 1060 | } |
| 1061 | |
| 1062 | template<typename Location> |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 1063 | void OnnxParserImpl::To1DTensor(const std::string& name, const Location& location) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1064 | { |
| 1065 | TensorShape shape = m_TensorsInfo[name].m_info->GetShape(); |
| 1066 | std::vector<uint32_t> newShape; |
| 1067 | for(uint i = 0; i < shape.GetNumDimensions() - 1; ++i) |
| 1068 | { |
| 1069 | if(shape[i] != 1) |
| 1070 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 1071 | throw ParseException( |
| 1072 | fmt::format("Only tensors with shape [1, ..., 1, X] can be converted to 1D and {} {}", |
| 1073 | TensorInfoAsString(*m_TensorsInfo[name].m_info, name, m_TensorsInfo[name].m_dtype), |
| 1074 | location.AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1075 | } |
| 1076 | } |
| 1077 | newShape.push_back(shape[shape.GetNumDimensions() - 1]); |
| 1078 | |
| 1079 | m_TensorsInfo[name].m_info->SetShape(TensorShape(static_cast<unsigned int>(newShape.size()), newShape.data())); |
| 1080 | } |
| 1081 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 1082 | void OnnxParserImpl::AddConvLayerWithDepthwiseConv(const onnx::NodeProto& node, const Convolution2dDescriptor& convDesc) |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1083 | { |
| 1084 | ARMNN_ASSERT(node.op_type() == "Conv"); |
| 1085 | |
| 1086 | DepthwiseConvolution2dDescriptor desc; |
| 1087 | desc.m_PadLeft = convDesc.m_PadLeft; |
| 1088 | desc.m_PadRight = convDesc.m_PadRight; |
| 1089 | desc.m_PadTop = convDesc.m_PadTop; |
| 1090 | desc.m_PadBottom = convDesc.m_PadBottom; |
| 1091 | desc.m_StrideX = convDesc.m_StrideX; |
| 1092 | desc.m_StrideY = convDesc.m_StrideY; |
| 1093 | desc.m_BiasEnabled = convDesc.m_BiasEnabled; |
| 1094 | |
Cathal Corbett | 0690265 | 2022-04-14 17:55:11 +0100 | [diff] [blame] | 1095 | armnn::IConnectableLayer* layer = m_Network->AddDepthwiseConvolution2dLayer(desc, node.name().c_str()); |
Cathal Corbett | 541880f | 2022-05-16 15:20:56 +0100 | [diff] [blame] | 1096 | std::string permuteStr = "permute_" + node.input(1); |
| 1097 | std::vector<std::string> tensorIndexes= {node.input(0), permuteStr}; |
Jan Eilers | 53ef795 | 2021-06-02 12:01:25 +0100 | [diff] [blame] | 1098 | |
Cathal Corbett | 541880f | 2022-05-16 15:20:56 +0100 | [diff] [blame] | 1099 | auto weightTensor = CreateConstTensor(node.input(1)); |
Cathal Corbett | 0690265 | 2022-04-14 17:55:11 +0100 | [diff] [blame] | 1100 | IConnectableLayer* weightsLayer = m_Network->AddConstantLayer(weightTensor.first); |
Cathal Corbett | 541880f | 2022-05-16 15:20:56 +0100 | [diff] [blame] | 1101 | |
| 1102 | // weights come in as [O,1,H,W] from ONNX and need to be converted to ArmNNs depthwise weights layout [1,H,W,O] |
| 1103 | armnn::PermutationVector perVec {3, 0, 1, 2}; |
| 1104 | TensorInfo weightsPermuted = armnnUtils::Permuted(weightTensor.first.GetInfo(), perVec); |
| 1105 | |
| 1106 | // Inserts NewLayer so layers don't need to be re-sorted. |
| 1107 | IConnectableLayer* permuteLayer = m_Network->AddPermuteLayer(PermuteDescriptor(perVec), |
| 1108 | "permute_layer"); |
| 1109 | permuteLayer->GetOutputSlot(0).SetTensorInfo(weightsPermuted); |
| 1110 | permuteLayer->GetOutputSlot(0).Connect(layer->GetInputSlot(1u)); |
| 1111 | |
Cathal Corbett | 0690265 | 2022-04-14 17:55:11 +0100 | [diff] [blame] | 1112 | weightsLayer->GetOutputSlot(0).SetTensorInfo(weightTensor.first.GetInfo()); |
Cathal Corbett | 541880f | 2022-05-16 15:20:56 +0100 | [diff] [blame] | 1113 | weightsLayer->GetOutputSlot(0).Connect(permuteLayer->GetInputSlot(0u)); |
Cathal Corbett | 0690265 | 2022-04-14 17:55:11 +0100 | [diff] [blame] | 1114 | |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1115 | if (node.input_size() == 3) |
| 1116 | { |
| 1117 | if(!m_TensorsInfo[node.input(2)].isConstant()) |
| 1118 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 1119 | throw ParseException(fmt::format("Bias '{}' should be constant in Conv layer '{}' {}", |
| 1120 | node.input(2), |
| 1121 | node.name(), |
| 1122 | CHECK_LOCATION().AsString())); |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1123 | } |
Cathal Corbett | 0690265 | 2022-04-14 17:55:11 +0100 | [diff] [blame] | 1124 | |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1125 | desc.m_BiasEnabled = true; |
| 1126 | auto biasTensor = CreateConstTensor(node.input(2)); |
Cathal Corbett | 0690265 | 2022-04-14 17:55:11 +0100 | [diff] [blame] | 1127 | tensorIndexes.emplace_back(node.input(2)); |
| 1128 | |
| 1129 | IConnectableLayer* biasLayer = m_Network->AddConstantLayer(biasTensor.first); |
| 1130 | biasLayer->GetOutputSlot(0).SetTensorInfo(biasTensor.first.GetInfo()); |
| 1131 | biasLayer->GetOutputSlot(0).Connect(layer->GetInputSlot(2u)); |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1132 | } |
Cathal Corbett | 0690265 | 2022-04-14 17:55:11 +0100 | [diff] [blame] | 1133 | |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1134 | ARMNN_ASSERT(layer != nullptr); |
| 1135 | |
| 1136 | auto outputInfo = ComputeOutputInfo({ node.output(0) }, layer, |
| 1137 | { m_TensorsInfo[node.input(0)].m_info->GetShape(), |
Cathal Corbett | 541880f | 2022-05-16 15:20:56 +0100 | [diff] [blame] | 1138 | weightsPermuted.GetShape() }); |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1139 | |
| 1140 | layer->GetOutputSlot(0).SetTensorInfo(outputInfo[0]); |
| 1141 | |
| 1142 | // register the input connection slots for the layer, connections are made after all layers have been created |
| 1143 | // only the tensors for the inputs are relevant, exclude the const tensors |
Cathal Corbett | 0690265 | 2022-04-14 17:55:11 +0100 | [diff] [blame] | 1144 | RegisterInputSlots(layer, tensorIndexes); |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1145 | |
| 1146 | // register the output connection slots for the layer, connections are made after all layers have been created |
| 1147 | RegisterOutputSlots(layer, {node.output(0)}); |
| 1148 | } |
| 1149 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 1150 | void OnnxParserImpl::AddFullyConnected(const onnx::NodeProto& matmulNode, const onnx::NodeProto* addNode) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1151 | { |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1152 | // find matmul inputs |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1153 | std::string inputName; |
Matthew Sloyan | ca36123 | 2023-02-16 14:50:22 +0000 | [diff] [blame] | 1154 | std::string weightName; |
| 1155 | std::string biasName; |
| 1156 | std::string outputName; |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1157 | CHECK_VALID_SIZE(static_cast<size_t>(matmulNode.input_size()), 2); |
| 1158 | CHECK_VALID_SIZE(static_cast<size_t>(matmulNode.output_size()), 1); |
| 1159 | VALID_INPUTS(matmulNode, STR_LIST(onnx::TensorProto::FLOAT)); |
| 1160 | |
| 1161 | GetInputAndParam(matmulNode, &inputName, &weightName, CHECK_LOCATION()); |
| 1162 | |
Matthew Sloyan | ca36123 | 2023-02-16 14:50:22 +0000 | [diff] [blame] | 1163 | TensorInfo inputInfo = *m_TensorsInfo[inputName].m_info; |
| 1164 | TensorInfo weightInfo = *m_TensorsInfo[weightName].m_info; |
| 1165 | TensorInfo biasInfo; |
| 1166 | |
| 1167 | std::vector<std::string> inputNames; |
| 1168 | |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1169 | FullyConnectedDescriptor desc; |
| 1170 | desc.m_BiasEnabled = addNode != nullptr; |
| 1171 | |
| 1172 | IConnectableLayer* layer = nullptr; |
| 1173 | if(desc.m_BiasEnabled) |
| 1174 | { |
| 1175 | // find bias const |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1176 | CHECK_VALID_SIZE(static_cast<size_t>(addNode->input_size()), 2); |
| 1177 | CHECK_VALID_SIZE(static_cast<size_t>(addNode->output_size()), 1); |
| 1178 | VALID_INPUTS(*addNode, STR_LIST(onnx::TensorProto::FLOAT)); |
| 1179 | |
| 1180 | GetInputAndParam(*addNode, nullptr, &biasName, CHECK_LOCATION()); |
| 1181 | |
| 1182 | //Output shape is [1, weights[1]] and 1d vec in ONNX can be [1,X] so we convert biases to "armnn" 1D |
| 1183 | To1DTensor(biasName, CHECK_LOCATION()); |
Matthew Sloyan | ca36123 | 2023-02-16 14:50:22 +0000 | [diff] [blame] | 1184 | biasInfo = *m_TensorsInfo[biasName].m_info; |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1185 | |
| 1186 | if (weightInfo.GetShape()[1] != biasInfo.GetShape()[0]) |
| 1187 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 1188 | throw ParseException( |
| 1189 | fmt::format("Shape of weights '{}' and bias of following Add node '{}' do not match : {}" |
| 1190 | " and {} ( /!\\ bias should be a 1D tensor) {}", |
| 1191 | weightName, |
| 1192 | addNode->name(), |
| 1193 | TensorInfoAsString(*m_TensorsInfo[weightName].m_info, weightName, |
| 1194 | m_TensorsInfo[weightName].m_dtype), |
| 1195 | TensorInfoAsString(*m_TensorsInfo[biasName].m_info, biasName, |
| 1196 | m_TensorsInfo[biasName].m_dtype ), |
| 1197 | CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1198 | } |
Matthew Sloyan | 81beae3 | 2021-07-13 19:46:11 +0100 | [diff] [blame] | 1199 | |
Matthew Sloyan | ca36123 | 2023-02-16 14:50:22 +0000 | [diff] [blame] | 1200 | inputNames = { inputName, weightName, biasName }; |
| 1201 | outputName = addNode->output(0); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1202 | } |
| 1203 | else |
| 1204 | { |
Matthew Sloyan | ca36123 | 2023-02-16 14:50:22 +0000 | [diff] [blame] | 1205 | inputNames = { inputName, weightName }; |
| 1206 | outputName = matmulNode.output(0); |
| 1207 | } |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1208 | |
Matthew Sloyan | ca36123 | 2023-02-16 14:50:22 +0000 | [diff] [blame] | 1209 | // Just add a FullyConnected layer, weights and biases are handled as inputs now. |
| 1210 | layer = m_Network->AddFullyConnectedLayer(desc, matmulNode.name().c_str()); |
| 1211 | ARMNN_ASSERT(layer != nullptr); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1212 | |
Matthew Sloyan | ca36123 | 2023-02-16 14:50:22 +0000 | [diff] [blame] | 1213 | if (inputInfo.GetNumDimensions() > 2) |
| 1214 | { |
| 1215 | // Add reshape to flatten to 2D [batch_size, input_size], |
| 1216 | // where "input_size" corresponds to the number of inputs to the layer, |
| 1217 | // matching the second dimension of weights, |
| 1218 | // and "batch_size" is calculated by dividing the number of elements by "input_size". |
| 1219 | std::vector<unsigned int> reshapedDimensions(2); |
| 1220 | reshapedDimensions[1] = weightInfo.GetShape()[0]; |
| 1221 | reshapedDimensions[0] = inputInfo.GetNumElements() / reshapedDimensions[1]; |
| 1222 | |
| 1223 | if (inputInfo.GetNumElements() % reshapedDimensions[1] != 0) |
Matthew Sloyan | 81beae3 | 2021-07-13 19:46:11 +0100 | [diff] [blame] | 1224 | { |
Matthew Sloyan | ca36123 | 2023-02-16 14:50:22 +0000 | [diff] [blame] | 1225 | throw ParseException( |
| 1226 | fmt::format("Failed to deduce input tensor shape from filter size {} {}", |
| 1227 | reshapedDimensions[1], |
| 1228 | CHECK_LOCATION().AsString())); |
Matthew Sloyan | 81beae3 | 2021-07-13 19:46:11 +0100 | [diff] [blame] | 1229 | } |
| 1230 | |
Matthew Sloyan | ca36123 | 2023-02-16 14:50:22 +0000 | [diff] [blame] | 1231 | TensorInfo reshapedTensorInfo = inputInfo; |
| 1232 | reshapedTensorInfo.SetShape(armnn::TensorShape{ 2, reshapedDimensions.data() }); |
| 1233 | inputInfo = reshapedTensorInfo; |
| 1234 | |
| 1235 | ReshapeDescriptor reshapeDescriptor; |
| 1236 | reshapeDescriptor.m_TargetShape = reshapedTensorInfo.GetShape(); |
| 1237 | |
| 1238 | std::string reshapeLayerName = fmt::format("Reshape_for:{}", layer->GetName()); |
| 1239 | IConnectableLayer* reshapeLayer = m_Network->AddReshapeLayer(reshapeDescriptor, reshapeLayerName.c_str()); |
| 1240 | |
| 1241 | reshapeLayer->GetOutputSlot(0).SetTensorInfo(reshapedTensorInfo); |
| 1242 | reshapeLayer->GetOutputSlot(0).Connect(layer->GetInputSlot(0)); |
| 1243 | |
| 1244 | RegisterInputSlots(reshapeLayer, {inputName}); |
| 1245 | inputNames[0] = reshapeLayerName; |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1246 | } |
Matthew Sloyan | ca36123 | 2023-02-16 14:50:22 +0000 | [diff] [blame] | 1247 | |
| 1248 | auto outputInfo = ComputeOutputInfo({ outputName }, |
| 1249 | layer, |
| 1250 | { inputInfo.GetShape(), |
| 1251 | weightInfo.GetShape() }); |
| 1252 | layer->GetOutputSlot(0).SetTensorInfo(outputInfo[0]); |
| 1253 | |
| 1254 | RegisterInputSlots(layer, inputNames); |
| 1255 | |
| 1256 | // Add constant layer to store weights/biases and connect to FullyConnected layer.. |
| 1257 | if(m_TensorsInfo[weightName].isConstant()) |
| 1258 | { |
| 1259 | IConnectableLayer* weightsLayer = m_Network->AddConstantLayer(CreateConstTensor(weightName).first); |
| 1260 | |
| 1261 | weightInfo.SetConstant(); |
| 1262 | weightsLayer->GetOutputSlot(0).SetTensorInfo(weightInfo); |
| 1263 | weightsLayer->GetOutputSlot(0).Connect(layer->GetInputSlot(1u)); |
| 1264 | } |
| 1265 | |
| 1266 | if(desc.m_BiasEnabled && m_TensorsInfo[biasName].isConstant()) |
| 1267 | { |
| 1268 | IConnectableLayer* biasLayer = m_Network->AddConstantLayer(CreateConstTensor(biasName).first); |
| 1269 | |
| 1270 | biasInfo.SetConstant(); |
| 1271 | biasLayer->GetOutputSlot(0).SetTensorInfo(biasInfo); |
| 1272 | biasLayer->GetOutputSlot(0).Connect(layer->GetInputSlot(2u)); |
| 1273 | } |
| 1274 | |
| 1275 | if (outputInfo[0].GetNumDimensions() > 2) |
| 1276 | { |
| 1277 | // Calculate reshape to flatten to 2D [batch_size, input_size] |
| 1278 | std::vector<unsigned int> reshapedDimensions(2); |
| 1279 | reshapedDimensions[1] = weightInfo.GetShape()[1]; |
| 1280 | reshapedDimensions[0] = outputInfo[0].GetNumElements() / reshapedDimensions[1]; |
| 1281 | |
| 1282 | if (outputInfo[0].GetNumElements() % reshapedDimensions[1] != 0) |
| 1283 | { |
| 1284 | throw ParseException( |
| 1285 | fmt::format("Failed to deduce output tensor shape from filter size {} {}", |
| 1286 | reshapedDimensions[1], |
| 1287 | CHECK_LOCATION().AsString())); |
| 1288 | } |
| 1289 | |
| 1290 | armnn::TensorInfo reshapedOutputTensorInfo = outputInfo[0]; |
| 1291 | reshapedOutputTensorInfo.SetShape(armnn::TensorShape{ 2, reshapedDimensions.data() }); |
| 1292 | layer->GetOutputSlot(0).SetTensorInfo(reshapedOutputTensorInfo); |
| 1293 | |
| 1294 | ReshapeDescriptor desc; |
| 1295 | desc.m_TargetShape = outputInfo[0].GetShape(); |
| 1296 | |
| 1297 | std::string reshapeLayerName = fmt::format("ExpandDims_for:{}", layer->GetName()); |
| 1298 | IConnectableLayer* reshapeLayer = m_Network->AddReshapeLayer(desc, reshapeLayerName.c_str()); |
| 1299 | |
| 1300 | layer->GetOutputSlot(0).Connect(reshapeLayer->GetInputSlot(0)); |
| 1301 | reshapeLayer->GetOutputSlot(0).SetTensorInfo(outputInfo[0]); |
| 1302 | |
| 1303 | RegisterInputSlots(reshapeLayer, {layer->GetName()}); |
| 1304 | layer = reshapeLayer; |
| 1305 | } |
| 1306 | |
| 1307 | RegisterOutputSlots(layer, { outputName }); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1308 | } |
| 1309 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 1310 | void OnnxParserImpl::AddPoolingLayer(const onnx::NodeProto& node, Pooling2dDescriptor& desc) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1311 | { |
| 1312 | |
| 1313 | CHECK_VALID_SIZE(static_cast<size_t>(node.input_size()), 1); |
| 1314 | CHECK_VALID_SIZE(static_cast<size_t>(node.output_size()), 1); |
| 1315 | |
| 1316 | VALID_INPUTS(node, STR_LIST(onnx::TensorProto::FLOAT)); |
| 1317 | |
| 1318 | std::vector<uint32_t> kernel_shape = ReadMandatoryNodeUint32ListAttribute(node, "kernel_shape"); //size of pool win |
| 1319 | std::vector<uint32_t> strides = ReadOptionalNodeUint32ListAttribute(node, "strides"); |
| 1320 | std::vector<uint32_t> pads = ReadOptionalNodeUint32ListAttribute(node, "pads"); |
| 1321 | |
| 1322 | desc.m_OutputShapeRounding = OutputShapeRounding::Floor; |
| 1323 | desc.m_PoolWidth = kernel_shape[1]; |
| 1324 | desc.m_PoolHeight = kernel_shape[0]; |
| 1325 | |
| 1326 | if(strides.empty()) |
| 1327 | { |
| 1328 | desc.m_StrideX = 1; |
| 1329 | desc.m_StrideY = 1; |
| 1330 | } |
| 1331 | else |
| 1332 | { |
| 1333 | desc.m_StrideX = strides[1]; |
| 1334 | desc.m_StrideY = strides[0]; |
| 1335 | } |
| 1336 | |
| 1337 | //Check new padding version first |
| 1338 | if(pads.empty()) |
| 1339 | { |
| 1340 | //Check deprecated version |
| 1341 | std::string paddingString = ReadOptionalNodeStringAttribute(node, "auto_pad"); |
| 1342 | if(paddingString != "VALID" && paddingString != "" && paddingString != "NOTSET") |
| 1343 | { |
| 1344 | bool isUpper; |
| 1345 | if( paddingString == "SAME_LOWER") |
| 1346 | { |
| 1347 | isUpper = false; |
| 1348 | } |
| 1349 | else if (paddingString == "SAME_UPPER") |
| 1350 | { |
| 1351 | isUpper = true; |
| 1352 | } |
| 1353 | else |
| 1354 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 1355 | throw ParseException(fmt::format("Invalid auto_pad attribute for node {}. " |
| 1356 | "Only SAME_UPPER, SAME_LOWER or VALID supported and found {} {}", |
| 1357 | node.name(), |
| 1358 | paddingString, |
| 1359 | CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1360 | } |
| 1361 | auto inputInfo = *m_TensorsInfo[node.input(0)].m_info; |
| 1362 | uint32_t inputHeight = inputInfo.GetShape()[2]; |
| 1363 | uint32_t inputWidth = inputInfo.GetShape()[3]; |
Sadik Armagan | 60bb9d8 | 2021-01-11 15:15:01 +0000 | [diff] [blame] | 1364 | CalcPadding(inputHeight, |
| 1365 | desc.m_PoolHeight, |
| 1366 | desc.m_StrideY, |
| 1367 | 1u, |
| 1368 | &desc.m_PadTop, |
| 1369 | &desc.m_PadBottom, |
| 1370 | isUpper); |
| 1371 | CalcPadding(inputWidth, |
| 1372 | desc.m_PoolWidth, |
| 1373 | desc.m_StrideX, |
| 1374 | 1u, |
| 1375 | &desc.m_PadLeft, |
| 1376 | &desc.m_PadRight, |
| 1377 | isUpper); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1378 | } |
| 1379 | } |
| 1380 | else |
| 1381 | { |
| 1382 | desc.m_PadTop = pads[0]; |
| 1383 | desc.m_PadLeft = pads[1]; |
| 1384 | desc.m_PadBottom = pads[2]; |
| 1385 | desc.m_PadRight = pads[3]; |
| 1386 | } |
| 1387 | |
| 1388 | IConnectableLayer* layer = m_Network->AddPooling2dLayer(desc, node.name().c_str()); |
Narumol Prangnawarat | ac2770a | 2020-04-01 16:51:23 +0100 | [diff] [blame] | 1389 | ARMNN_ASSERT(layer != nullptr); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1390 | |
| 1391 | auto outputInfo = ComputeOutputInfo({node.output(0)}, layer, {m_TensorsInfo[node.input(0)].m_info->GetShape()}); |
| 1392 | layer->GetOutputSlot(0).SetTensorInfo(outputInfo[0]); |
| 1393 | |
| 1394 | // register the input connection slots for the layer, connections are made after all layers have been created |
| 1395 | // only the tensors for the inputs are relevant, exclude the const tensors |
| 1396 | RegisterInputSlots(layer, {node.input(0)}); |
| 1397 | |
| 1398 | // register the output connection slots for the layer, connections are made after all layers have been created |
| 1399 | RegisterOutputSlots(layer, {node.output(0)}); |
| 1400 | } |
| 1401 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 1402 | std::pair<std::string, std::string> OnnxParserImpl::AddPrepareBroadcast(const std::string& input0, |
| 1403 | const std::string& input1) |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1404 | { |
| 1405 | std::pair<std::string, std::string> inputs = std::make_pair(input0, input1); |
| 1406 | |
| 1407 | TensorShape input0Shape = m_TensorsInfo[input0].m_info->GetShape(); |
| 1408 | TensorShape input1Shape = m_TensorsInfo[input1].m_info->GetShape(); |
| 1409 | |
| 1410 | if(input1Shape.GetNumDimensions() < input0Shape.GetNumDimensions()) |
| 1411 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 1412 | auto outputName = fmt::format("reshape_output_{}", input1); |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1413 | PrependForBroadcast(outputName, input1, input0); |
| 1414 | inputs.second = outputName; |
| 1415 | } |
| 1416 | else if(input0Shape.GetNumDimensions() < input1Shape.GetNumDimensions()) |
| 1417 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 1418 | auto outputName = fmt::format("reshape_output_{}", input0); |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1419 | PrependForBroadcast(outputName, input0, input1); |
| 1420 | inputs.first = outputName; |
| 1421 | } |
| 1422 | return inputs; |
| 1423 | } |
| 1424 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 1425 | void OnnxParserImpl::CreateConstantLayer(const std::string& tensorName, const std::string& layerName) |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1426 | { |
| 1427 | auto armnnTensor = CreateConstTensor(tensorName); |
Narumol Prangnawarat | f10b15a | 2021-09-17 21:08:57 +0100 | [diff] [blame] | 1428 | IConnectableLayer* layer = m_Network->AddConstantLayer(armnnTensor.first, layerName.c_str()); |
| 1429 | layer->GetOutputSlot(0).SetTensorInfo(armnnTensor.first.GetInfo()); |
| 1430 | RegisterOutputSlots(layer, {tensorName}); |
| 1431 | } |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1432 | |
Narumol Prangnawarat | f10b15a | 2021-09-17 21:08:57 +0100 | [diff] [blame] | 1433 | void OnnxParserImpl::CreateInt64ConstantLayer(const std::string& tensorName, const std::string& layerName) |
| 1434 | { |
| 1435 | auto armnnTensor = CreateInt64ConstTensor(tensorName); |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1436 | IConnectableLayer* layer = m_Network->AddConstantLayer(armnnTensor.first, layerName.c_str()); |
| 1437 | layer->GetOutputSlot(0).SetTensorInfo(armnnTensor.first.GetInfo()); |
| 1438 | RegisterOutputSlots(layer, {tensorName}); |
| 1439 | } |
| 1440 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 1441 | void OnnxParserImpl::CreateReshapeLayer(const std::string& inputName, |
| 1442 | const std::string& outputName, |
| 1443 | const std::string& layerName) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1444 | { |
| 1445 | const TensorInfo outputTensorInfo = *m_TensorsInfo[outputName].m_info; |
| 1446 | ReshapeDescriptor reshapeDesc; |
| 1447 | reshapeDesc.m_TargetShape = outputTensorInfo.GetShape(); |
| 1448 | |
| 1449 | IConnectableLayer* layer = m_Network->AddReshapeLayer(reshapeDesc, layerName.c_str()); |
Narumol Prangnawarat | ac2770a | 2020-04-01 16:51:23 +0100 | [diff] [blame] | 1450 | ARMNN_ASSERT(layer != nullptr); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1451 | layer->GetOutputSlot(0).SetTensorInfo(outputTensorInfo); |
| 1452 | |
| 1453 | // register the input connection slots for the layer, connections are made after all layers have been created |
| 1454 | // only the tensors for the inputs are relevant, exclude the const tensors |
| 1455 | RegisterInputSlots(layer, {inputName}); |
| 1456 | |
| 1457 | // register the output connection slots for the layer, connections are made after all layers have been created |
| 1458 | RegisterOutputSlots(layer, {outputName}); |
| 1459 | } |
| 1460 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 1461 | void OnnxParserImpl::ParseActivation(const onnx::NodeProto& node, const armnn::ActivationFunction func) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1462 | { |
Finn Williams | 7ee5d2c | 2020-03-27 11:11:50 +0000 | [diff] [blame] | 1463 | CHECK_VALID_SIZE(static_cast<size_t>(node.input_size()), 1, 3); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1464 | CHECK_VALID_SIZE(static_cast<size_t>(node.output_size()), 1); |
| 1465 | |
| 1466 | VALID_INPUTS(node, STR_LIST(onnx::TensorProto::FLOAT)); |
| 1467 | |
| 1468 | ActivationDescriptor desc; |
Tee Jung | 7ff9a60 | 2019-11-01 07:04:42 +0000 | [diff] [blame] | 1469 | desc.m_Function = func; |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1470 | |
Finn Williams | 7ee5d2c | 2020-03-27 11:11:50 +0000 | [diff] [blame] | 1471 | if (func == ActivationFunction::BoundedReLu) |
| 1472 | { |
Narumol Prangnawarat | f106ab7 | 2021-09-15 17:30:37 +0100 | [diff] [blame] | 1473 | if (node.input_size() == 1 && node.attribute_size() > 0) |
| 1474 | { |
| 1475 | desc.m_A = ReadOptionalNodeFloatAttribute(node, "max", std::numeric_limits<float>::max()); |
| 1476 | desc.m_B = ReadOptionalNodeFloatAttribute(node, "min", std::numeric_limits<float>::lowest()); |
| 1477 | } |
| 1478 | else |
| 1479 | { |
| 1480 | desc.m_A = node.input(2).empty() ? std::numeric_limits<float>::max() : std::stof(node.input(2)); |
| 1481 | desc.m_B = node.input(1).empty() ? std::numeric_limits<float>::lowest() : std::stof(node.input(1)); |
| 1482 | } |
Finn Williams | 7ee5d2c | 2020-03-27 11:11:50 +0000 | [diff] [blame] | 1483 | } |
| 1484 | |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1485 | IConnectableLayer* const layer = m_Network->AddActivationLayer(desc, node.name().c_str()); |
Narumol Prangnawarat | ac2770a | 2020-04-01 16:51:23 +0100 | [diff] [blame] | 1486 | ARMNN_ASSERT(layer != nullptr); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1487 | |
| 1488 | auto outputInfo = ComputeOutputInfo({ node.output(0)}, layer, {m_TensorsInfo[node.input(0)].m_info->GetShape()}); |
| 1489 | layer->GetOutputSlot(0).SetTensorInfo(outputInfo[0]); |
| 1490 | |
| 1491 | // register the input connection slots for the layer, connections are made after all layers have been created |
| 1492 | // only the tensors for the inputs are relevant, exclude the const tensors |
| 1493 | RegisterInputSlots(layer, {node.input(0)}); |
| 1494 | |
| 1495 | // register the output connection slots for the layer, connections are made after all layers have been created |
| 1496 | RegisterOutputSlots(layer, {node.output(0)}); |
| 1497 | } |
| 1498 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 1499 | void OnnxParserImpl::ParseClip(const onnx::NodeProto& node) |
Finn Williams | 7ee5d2c | 2020-03-27 11:11:50 +0000 | [diff] [blame] | 1500 | { |
| 1501 | ParseActivation(node, ActivationFunction::BoundedReLu); |
| 1502 | } |
| 1503 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 1504 | void OnnxParserImpl::ParseSigmoid(const onnx::NodeProto& node) |
Tee Jung | 7ff9a60 | 2019-11-01 07:04:42 +0000 | [diff] [blame] | 1505 | { |
| 1506 | ParseActivation(node, ActivationFunction::Sigmoid); |
| 1507 | } |
| 1508 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 1509 | void OnnxParserImpl::ParseTanh(const onnx::NodeProto& node) |
Tee Jung | 7ff9a60 | 2019-11-01 07:04:42 +0000 | [diff] [blame] | 1510 | { |
| 1511 | ParseActivation(node, ActivationFunction::TanH); |
| 1512 | } |
| 1513 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 1514 | void OnnxParserImpl::ParseRelu(const onnx::NodeProto& node) |
Tee Jung | 7ff9a60 | 2019-11-01 07:04:42 +0000 | [diff] [blame] | 1515 | { |
| 1516 | ParseActivation(node, ActivationFunction::ReLu); |
| 1517 | } |
| 1518 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 1519 | void OnnxParserImpl::ParseLeakyRelu(const onnx::NodeProto& node) |
Tee Jung | 7ff9a60 | 2019-11-01 07:04:42 +0000 | [diff] [blame] | 1520 | { |
| 1521 | ParseActivation(node, ActivationFunction::LeakyReLu); |
| 1522 | } |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1523 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 1524 | void OnnxParserImpl::ParseAdd(const onnx::NodeProto& node) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1525 | { |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1526 | CHECK_VALID_SIZE(static_cast<size_t>(node.input_size()), 2); |
| 1527 | CHECK_VALID_SIZE(static_cast<size_t>(node.output_size()), 1); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1528 | |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1529 | VALID_INPUTS(node, STR_LIST(onnx::TensorProto::FLOAT)); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1530 | |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1531 | // TODO: unify broadcast validation code across layers |
| 1532 | // tracked by: IVGCVSW-1576 |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1533 | |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1534 | // Checking broadcast compatibility : only scalar or 1D tensors |
| 1535 | auto inputs = AddPrepareBroadcast(node.input(0), node.input(1)); |
| 1536 | auto input0 = *m_TensorsInfo[inputs.first].m_info; |
| 1537 | auto input1 = *m_TensorsInfo[inputs.second].m_info; |
| 1538 | ARMNN_ASSERT(input0.GetNumDimensions() == input1.GetNumDimensions()); |
| 1539 | |
| 1540 | unsigned int numDims = input0.GetNumDimensions(); |
| 1541 | for (unsigned int i = 0; i < numDims; i++) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1542 | { |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1543 | unsigned int dim0 = input0.GetShape()[i]; |
| 1544 | unsigned int dim1 = input1.GetShape()[i]; |
| 1545 | if (dim0 != dim1 && dim0 != 1 && dim1 != 1) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1546 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 1547 | throw ParseException( |
| 1548 | fmt::format("Broadcast is only supported for scalar or 1D tensors in Add node '{}'. " |
| 1549 | "Input dimensions should either match or one should be of size 1 and here, " |
| 1550 | "{} and {} {}", |
| 1551 | node.name(), |
| 1552 | TensorInfoAsString(*m_TensorsInfo[inputs.first].m_info, inputs.first, |
| 1553 | m_TensorsInfo[inputs.first].m_dtype), |
| 1554 | TensorInfoAsString(*m_TensorsInfo[inputs.second].m_info, inputs.second, |
| 1555 | m_TensorsInfo[inputs.second].m_dtype), |
| 1556 | CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1557 | } |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1558 | } |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1559 | |
| 1560 | |
| 1561 | IConnectableLayer* layer = m_Network->AddAdditionLayer(node.name().c_str()); |
Narumol Prangnawarat | ac2770a | 2020-04-01 16:51:23 +0100 | [diff] [blame] | 1562 | ARMNN_ASSERT(layer != nullptr); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1563 | |
| 1564 | auto outputInfo = ComputeOutputInfo({ node.output(0) }, layer, |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1565 | { m_TensorsInfo[inputs.first].m_info->GetShape(), |
| 1566 | m_TensorsInfo[inputs.second].m_info->GetShape() }); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1567 | layer->GetOutputSlot(0).SetTensorInfo(outputInfo[0]); |
| 1568 | |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1569 | // register the input connection -> for constant inputs, we need to make a newDim constant layer |
| 1570 | if(m_TensorsInfo[inputs.first].isConstant()) { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 1571 | CreateConstantLayer(inputs.first, fmt::format("Add:constant_of_{}", node.input(0))); |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1572 | } |
| 1573 | if(m_TensorsInfo[inputs.second].isConstant()) { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 1574 | CreateConstantLayer(inputs.second, fmt::format("Add:constant_of_{}", node.input(1))); |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1575 | } |
| 1576 | RegisterInputSlots(layer, {inputs.first, inputs.second}); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1577 | |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1578 | // register the output connection |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1579 | RegisterOutputSlots(layer, {node.output(0)}); |
| 1580 | } |
| 1581 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 1582 | void OnnxParserImpl::ParseAveragePool(const onnx::NodeProto& node) |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1583 | { |
| 1584 | Pooling2dDescriptor desc; |
| 1585 | desc.m_PoolType = PoolingAlgorithm::Average; |
| 1586 | |
| 1587 | uint32_t count_include_pad = 0; |
| 1588 | count_include_pad = ReadOptionalNodeUint32Attribute(node, "count_include_pad"); |
| 1589 | if(count_include_pad) { |
| 1590 | desc.m_PaddingMethod = PaddingMethod::IgnoreValue; |
| 1591 | } |
| 1592 | AddPoolingLayer(node, desc); |
| 1593 | } |
| 1594 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 1595 | void OnnxParserImpl::ParseBatchNormalization(const onnx::NodeProto& node) |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1596 | { |
| 1597 | //IGNORE momentum parameter and spatial parameters |
| 1598 | |
| 1599 | CHECK_VALID_SIZE(static_cast<size_t>(node.input_size()), 5); |
| 1600 | CHECK_VALID_SIZE(static_cast<size_t>(node.output_size()), 1); |
| 1601 | |
| 1602 | VALID_INPUTS(node, STR_LIST(onnx::TensorProto::FLOAT)); |
| 1603 | for(int ind = 1; ind < node.input_size(); ++ind) |
| 1604 | { |
| 1605 | auto tensor = node.input(ind); |
| 1606 | if(! m_TensorsInfo[tensor].isConstant()) |
| 1607 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 1608 | throw ParseException( |
| 1609 | fmt::format("Input tensor '{}' should be constant in BatchNormalization node '{}' {}", |
| 1610 | tensor, |
| 1611 | node.name(), |
| 1612 | CHECK_LOCATION().AsString())); |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1613 | } |
| 1614 | } |
| 1615 | |
| 1616 | float epsilon = ReadOptionalNodeFloatAttribute(node, "epsilon", 1e-5f); |
| 1617 | BatchNormalizationDescriptor desc; |
| 1618 | desc.m_Eps = epsilon; |
| 1619 | |
| 1620 | auto scaleTensor = CreateConstTensor(node.input(1)); |
| 1621 | auto biasTensor = CreateConstTensor(node.input(2)); |
| 1622 | auto meanTensor = CreateConstTensor(node.input(3)); |
| 1623 | auto varTensor = CreateConstTensor(node.input(4)); |
| 1624 | |
| 1625 | IConnectableLayer* layer = m_Network->AddBatchNormalizationLayer(desc, |
| 1626 | meanTensor.first, |
| 1627 | varTensor.first, |
| 1628 | biasTensor.first, |
| 1629 | scaleTensor.first, |
| 1630 | node.name().c_str()); |
| 1631 | ARMNN_ASSERT(layer != nullptr); |
| 1632 | |
| 1633 | auto outputInfo = ComputeOutputInfo({node.output(0)}, layer, {m_TensorsInfo[node.input(0)].m_info->GetShape()}); |
| 1634 | layer->GetOutputSlot(0).SetTensorInfo(outputInfo[0]); |
| 1635 | |
| 1636 | RegisterInputSlots(layer, {node.input(0)}); //don't register constant inputs |
| 1637 | |
| 1638 | // register the output connection |
| 1639 | RegisterOutputSlots(layer, {node.output(0)}); |
| 1640 | } |
| 1641 | |
Narumol Prangnawarat | bc3bb62 | 2021-09-24 16:08:34 +0100 | [diff] [blame] | 1642 | void OnnxParserImpl::ParseConcat(const onnx::NodeProto& node) |
| 1643 | { |
| 1644 | CHECK_VALID_SIZE(static_cast<size_t>(node.output_size()), 1); |
| 1645 | |
| 1646 | uint32_t numConcatView = static_cast<uint32_t>(node.input_size()); |
| 1647 | uint32_t inputRank = m_TensorsInfo[node.input(0)].m_info->GetNumDimensions(); |
| 1648 | |
| 1649 | int axisInt = ReadMandatoryNodeIntAttribute(node, "axis"); |
| 1650 | |
| 1651 | unsigned int concatDimInput = static_cast<unsigned int>( |
| 1652 | (static_cast<int>(inputRank) + axisInt) % static_cast<int>(inputRank)); |
| 1653 | |
| 1654 | OriginsDescriptor concatDescriptor(numConcatView, inputRank); |
| 1655 | concatDescriptor.SetConcatAxis(concatDimInput); |
| 1656 | |
| 1657 | unsigned int mergeDimOrigin = 0; |
| 1658 | |
| 1659 | std::vector<TensorShape> inputShapes; |
| 1660 | std::vector<std::string> tensorIds; |
| 1661 | |
| 1662 | for (unsigned int viewIndex = 0; viewIndex < numConcatView; ++viewIndex) |
| 1663 | { |
| 1664 | std::string nodeName = node.input(static_cast<int>(viewIndex)); |
| 1665 | auto inputTensorInfo = *m_TensorsInfo[nodeName].m_info; |
| 1666 | inputShapes.push_back(inputTensorInfo.GetShape()); |
| 1667 | tensorIds.push_back(nodeName); |
| 1668 | |
| 1669 | // Set up concatDescriptor view origin |
| 1670 | armnnUtils::ProcessConcatInputTensorInfo( |
| 1671 | inputTensorInfo, concatDescriptor, concatDimInput, viewIndex, mergeDimOrigin); |
| 1672 | } |
| 1673 | |
| 1674 | IConnectableLayer* layer = m_Network->AddConcatLayer(concatDescriptor, node.name().c_str()); |
| 1675 | ARMNN_ASSERT(layer != nullptr); |
| 1676 | |
Narumol Prangnawarat | 452274c | 2021-09-23 16:12:19 +0100 | [diff] [blame] | 1677 | auto outputInfo = ComputeOutputInfo({node.output(0)}, layer, inputShapes, |
| 1678 | m_TensorsInfo[node.input(0)].m_dtype); |
Narumol Prangnawarat | bc3bb62 | 2021-09-24 16:08:34 +0100 | [diff] [blame] | 1679 | |
| 1680 | layer->GetOutputSlot(0).SetTensorInfo(outputInfo[0]); |
| 1681 | |
| 1682 | // register the input connection slots for the layer, connections are made after all layers have been created |
| 1683 | RegisterInputSlots(layer, tensorIds); |
| 1684 | |
| 1685 | // register the output connection slots for the layer, connections are made after all layers have been created |
| 1686 | RegisterOutputSlots(layer, { node.output(0) }); |
| 1687 | } |
| 1688 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 1689 | void OnnxParserImpl::ParseConstant(const onnx::NodeProto& node) |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1690 | { |
| 1691 | CHECK_VALID_SIZE(static_cast<size_t>(node.attribute_size()), 1); |
| 1692 | if (!node.attribute(0).has_t()) |
| 1693 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 1694 | throw ParseException(fmt::format("Value not found for Constant node '{}' {}", |
| 1695 | node.name(), |
| 1696 | CHECK_LOCATION().AsString())); |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1697 | } |
| 1698 | const onnx::TensorProto& onnxTensor = node.attribute(0).t(); |
| 1699 | |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1700 | //Register this as a m_ConstParam so we know we can use it as a constant param in future layers. |
| 1701 | m_TensorsInfo[node.output(0)].m_tensor = std::make_unique<const onnx::TensorProto>(onnxTensor); |
| 1702 | m_TensorsInfo[node.output(0)].m_info = std::make_unique<TensorInfo>(ToTensorInfo(onnxTensor)); |
| 1703 | m_TensorsInfo[node.output(0)].m_dtype = static_cast<onnx::TensorProto::DataType>(onnxTensor.data_type()); |
| 1704 | |
Narumol Prangnawarat | f10b15a | 2021-09-17 21:08:57 +0100 | [diff] [blame] | 1705 | if (m_TensorsInfo[node.output(0)].m_dtype == onnx::TensorProto_DataType_FLOAT) |
| 1706 | { |
| 1707 | CreateConstantLayer(node.output(0), node.name()); |
| 1708 | } |
| 1709 | else if (m_TensorsInfo[node.output(0)].m_dtype == onnx::TensorProto_DataType_INT64) |
| 1710 | { |
| 1711 | CreateInt64ConstantLayer(node.output(0), node.name()); |
| 1712 | } |
| 1713 | else |
| 1714 | { |
| 1715 | throw ParseException(fmt::format("Data type not support for Constant node '{}' {}", |
| 1716 | node.name(), |
| 1717 | CHECK_LOCATION().AsString())); |
| 1718 | } |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1719 | } |
| 1720 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 1721 | void OnnxParserImpl::ParseConv(const onnx::NodeProto& node) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1722 | { |
| 1723 | CHECK_VALID_SIZE(static_cast<size_t>(node.input_size()), 2, 3); //input, weight, (bias) |
| 1724 | CHECK_VALID_SIZE(static_cast<size_t>(node.output_size()), 1); |
| 1725 | |
| 1726 | VALID_INPUTS(node, STR_LIST(onnx::TensorProto::FLOAT)); |
| 1727 | |
| 1728 | if(m_TensorsInfo[node.input(0)].m_info->GetNumDimensions() != 4) |
| 1729 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 1730 | throw ParseException( |
| 1731 | fmt::format("ArmNN only supports 2D convolution and Conv layer '{}' input {} {}", |
| 1732 | node.name(), |
| 1733 | TensorInfoAsString(*m_TensorsInfo[node.input(0)].m_info, node.input(0), |
| 1734 | m_TensorsInfo[node.input(0)].m_dtype), |
| 1735 | CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1736 | } |
| 1737 | |
| 1738 | if(!m_TensorsInfo[node.input(1)].isConstant()) |
| 1739 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 1740 | throw ParseException( |
| 1741 | fmt::format("Weights '{}' should be constant in Conv layer '{}' {}", |
| 1742 | node.input(1), |
| 1743 | node.name(), |
| 1744 | CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1745 | } |
| 1746 | |
| 1747 | auto inputInfo = *m_TensorsInfo[node.input(0)].m_info; |
| 1748 | |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1749 | Convolution2dDescriptor desc; |
| 1750 | desc.m_BiasEnabled = false; |
| 1751 | |
| 1752 | std::vector<uint32_t> strides = ReadOptionalNodeUint32ListAttribute(node, "strides"); |
| 1753 | if(strides.empty()) |
| 1754 | { |
| 1755 | desc.m_StrideX = 1; |
| 1756 | desc.m_StrideY = 1; |
| 1757 | } |
| 1758 | else |
| 1759 | { |
| 1760 | desc.m_StrideX = strides[1]; |
| 1761 | desc.m_StrideY = strides[0]; |
| 1762 | } |
| 1763 | |
Sadik Armagan | 60bb9d8 | 2021-01-11 15:15:01 +0000 | [diff] [blame] | 1764 | std::vector<uint32_t> dilations = ReadOptionalNodeUint32ListAttribute(node, "dilations"); |
| 1765 | if(!dilations.empty()) |
| 1766 | { |
| 1767 | desc.m_DilationX = dilations[1]; |
| 1768 | desc.m_DilationY = dilations[0]; |
| 1769 | } |
| 1770 | |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1771 | std::vector<uint32_t> pads = ReadOptionalNodeUint32ListAttribute(node, "pads"); |
| 1772 | //Check new padding version first |
| 1773 | if(pads.empty()) |
| 1774 | { |
| 1775 | //Check deprecated version |
| 1776 | std::string paddingString = ReadOptionalNodeStringAttribute(node, "auto_pad"); |
| 1777 | if(paddingString != "VALID" && paddingString != "" && paddingString != "NOTSET") |
| 1778 | { |
| 1779 | bool isUpper; |
| 1780 | if( paddingString == "SAME_LOWER") |
| 1781 | { |
| 1782 | isUpper = false; |
| 1783 | } |
| 1784 | else if (paddingString == "SAME_UPPER") |
| 1785 | { |
| 1786 | isUpper = true; |
| 1787 | } |
| 1788 | else |
| 1789 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 1790 | throw ParseException( |
| 1791 | fmt::format("Invalid auto_pad attribute for node {}. Only SAME_UPPER, SAME_LOWER or VALID " |
| 1792 | "supported and found {} {}", |
| 1793 | node.name(), |
| 1794 | paddingString, |
| 1795 | CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1796 | } |
| 1797 | uint32_t inputHeight = inputInfo.GetShape()[2]; |
| 1798 | uint32_t inputWidth = inputInfo.GetShape()[3]; |
| 1799 | |
| 1800 | uint32_t weightHeight; |
| 1801 | uint32_t weightWidth; |
| 1802 | std::vector<uint32_t> kernel_shape = ReadOptionalNodeUint32ListAttribute(node, "kernel_shape"); |
| 1803 | if (kernel_shape.empty()) |
| 1804 | { |
| 1805 | const TensorInfo weightTensorInfo = *m_TensorsInfo[node.input(1)].m_info; |
| 1806 | weightHeight = weightTensorInfo.GetShape()[2]; |
| 1807 | weightWidth = weightTensorInfo.GetShape()[3]; |
| 1808 | } |
| 1809 | else |
| 1810 | { |
| 1811 | weightHeight = kernel_shape[0]; |
| 1812 | weightWidth = kernel_shape[1]; |
| 1813 | } |
Sadik Armagan | 60bb9d8 | 2021-01-11 15:15:01 +0000 | [diff] [blame] | 1814 | CalcPadding(inputHeight, |
| 1815 | weightHeight, |
| 1816 | desc.m_StrideY, |
| 1817 | desc.m_DilationY, |
| 1818 | &desc.m_PadTop, |
| 1819 | &desc.m_PadBottom, |
| 1820 | isUpper); |
| 1821 | CalcPadding(inputWidth, |
| 1822 | weightWidth, |
| 1823 | desc.m_StrideX, |
| 1824 | desc.m_DilationX, |
| 1825 | &desc.m_PadLeft, |
| 1826 | &desc.m_PadRight, |
| 1827 | isUpper); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1828 | } |
| 1829 | } |
| 1830 | else |
| 1831 | { |
| 1832 | desc.m_PadTop = pads[0]; |
| 1833 | desc.m_PadLeft = pads[1]; |
| 1834 | desc.m_PadBottom = pads[2]; |
| 1835 | desc.m_PadRight = pads[3]; |
| 1836 | } |
| 1837 | |
| 1838 | uint32_t group = ReadOptionalNodeUint32Attribute(node, "group", 1); |
| 1839 | if(group > 1) |
| 1840 | { |
| 1841 | if (group > inputInfo.GetShape()[1]) |
| 1842 | { |
| 1843 | throw ParseException( |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 1844 | fmt::format("Error parsing Convolution node: {}. " |
| 1845 | "The 'group'={} parameter cannot be larger than the " |
| 1846 | "channel of the input shape={} (in NCHW format). {}", |
| 1847 | node.name(), |
| 1848 | group, |
| 1849 | inputInfo.GetShape()[1], |
| 1850 | CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1851 | } |
| 1852 | else if (group == inputInfo.GetShape()[1]) |
| 1853 | { |
| 1854 | // we use a depthwise convolution here, because the number of groups equals to the |
| 1855 | // input channels |
| 1856 | AddConvLayerWithDepthwiseConv(node, desc); |
| 1857 | return; |
| 1858 | } |
| 1859 | else |
| 1860 | { |
| 1861 | // TODO: split the input by channels into channels/groups separate convolutions |
Jim Flynn | e242f2d | 2019-05-22 14:24:13 +0100 | [diff] [blame] | 1862 | // and concatenate the results afterwards |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 1863 | throw ParseException(fmt::format("Error parsing Convolution node: {}. " |
| 1864 | "The 'group'={} parameter should be 1 or be equal to the " |
| 1865 | "channel of the input shape={} (in NCHW format). {}", |
| 1866 | node.name(), |
| 1867 | group, |
| 1868 | inputInfo.GetShape()[1], |
| 1869 | CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1870 | } |
| 1871 | } |
| 1872 | |
Keith Davis | 721e629 | 2022-05-17 10:06:53 +0100 | [diff] [blame] | 1873 | node.input_size() == 3 ? desc.m_BiasEnabled = true : desc.m_BiasEnabled = false; |
| 1874 | armnn::IConnectableLayer* layer = m_Network->AddConvolution2dLayer(desc, node.name().c_str()); |
Keith Davis | b4dd5cc | 2022-04-07 11:32:00 +0100 | [diff] [blame] | 1875 | std::vector<std::string> tensorIndexes= {node.input(0), node.input(1)}; |
| 1876 | |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1877 | auto weightTensor = CreateConstTensor(node.input(1)); |
| 1878 | |
Keith Davis | 721e629 | 2022-05-17 10:06:53 +0100 | [diff] [blame] | 1879 | IConnectableLayer* weightsLayer = m_Network->AddConstantLayer(weightTensor.first); |
| 1880 | weightsLayer->GetOutputSlot(0).SetTensorInfo(weightTensor.first.GetInfo()); |
| 1881 | weightsLayer->GetOutputSlot(0).Connect(layer->GetInputSlot(1u)); |
| 1882 | |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1883 | if (node.input_size() == 3) |
| 1884 | { |
| 1885 | if(!m_TensorsInfo[node.input(2)].isConstant()) |
| 1886 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 1887 | throw ParseException(fmt::format("Bias '{}' should be constant in Conv layer '{}' {}", |
| 1888 | node.input(2), |
| 1889 | node.name(), |
| 1890 | CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1891 | } |
| 1892 | desc.m_BiasEnabled = true; |
| 1893 | auto biasTensor = CreateConstTensor(node.input(2)); |
Keith Davis | 721e629 | 2022-05-17 10:06:53 +0100 | [diff] [blame] | 1894 | |
| 1895 | IConnectableLayer* biasLayer = m_Network->AddConstantLayer(biasTensor.first); |
| 1896 | biasLayer->GetOutputSlot(0).SetTensorInfo(biasTensor.first.GetInfo()); |
| 1897 | biasLayer->GetOutputSlot(0).Connect(layer->GetInputSlot(2u)); |
| 1898 | |
| 1899 | tensorIndexes.emplace_back(node.input(2)); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1900 | } |
Keith Davis | 721e629 | 2022-05-17 10:06:53 +0100 | [diff] [blame] | 1901 | |
Narumol Prangnawarat | ac2770a | 2020-04-01 16:51:23 +0100 | [diff] [blame] | 1902 | ARMNN_ASSERT(layer != nullptr); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1903 | |
| 1904 | auto outputInfo = ComputeOutputInfo({ node.output(0) }, layer, |
| 1905 | { m_TensorsInfo[node.input(0)].m_info->GetShape(), |
| 1906 | m_TensorsInfo[node.input(1)].m_info->GetShape() }); |
| 1907 | layer->GetOutputSlot(0).SetTensorInfo(outputInfo[0]); |
| 1908 | |
| 1909 | // register the input connection slots for the layer, connections are made after all layers have been created |
| 1910 | // only the tensors for the inputs are relevant, exclude the const tensors |
Keith Davis | b4dd5cc | 2022-04-07 11:32:00 +0100 | [diff] [blame] | 1911 | RegisterInputSlots(layer, tensorIndexes); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 1912 | |
| 1913 | // register the output connection slots for the layer, connections are made after all layers have been created |
| 1914 | RegisterOutputSlots(layer, {node.output(0)}); |
| 1915 | } |
| 1916 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 1917 | void OnnxParserImpl::ParseFlatten(const onnx::NodeProto& node) |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1918 | { |
| 1919 | CHECK_VALID_SIZE(static_cast<size_t>(node.input_size()), 1); |
| 1920 | CHECK_VALID_SIZE(static_cast<size_t>(node.output_size()), 1); |
| 1921 | |
| 1922 | CHECK_VALID_DATATYPE(node.name(), node.input(0), |
| 1923 | m_TensorsInfo[node.input(0)].m_dtype, |
| 1924 | onnx::TensorProto::FLOAT); |
| 1925 | |
| 1926 | int64_t axis = ReadOptionalNodeInt64Attribute(node, "axis", 1); |
| 1927 | TensorShape inputShape = m_TensorsInfo[node.input(0)].m_info->GetShape(); |
| 1928 | |
| 1929 | /// Negative axis conversion |
| 1930 | if (axis < 0) |
| 1931 | { |
| 1932 | axis += inputShape.GetNumDimensions(); |
| 1933 | } |
| 1934 | |
| 1935 | /// Check Axis is within dimensions |
| 1936 | if (axis < 0 || axis >= inputShape.GetNumDimensions()) |
| 1937 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 1938 | throw ParseException(fmt::format("Axis '{}' invalid. Tensor has '{}' dimensions in FlattenLayer '{}'", |
| 1939 | axis, inputShape.GetNumDimensions(), node.name())); |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 1940 | } |
| 1941 | |
| 1942 | /// If axis chosen is 0 dimension1 will always be 1 in output , default dimension2 to 1 because 0 is invalid |
| 1943 | uint dimension1{1}; |
| 1944 | uint dimension2{1}; |
| 1945 | uint i{0}; |
| 1946 | |
| 1947 | /// dimension1 = (d_0 * d_1 ... d_(axis-1)) |
| 1948 | for (i = 0; i < axis; i++){ |
| 1949 | dimension1 *= inputShape[i]; |
| 1950 | } |
| 1951 | |
| 1952 | /// dimension2 = (d_axis * d_(axis+1) ... d_n) |
| 1953 | for (i = static_cast<uint>(axis); i < inputShape.GetNumDimensions(); i++){ |
| 1954 | dimension2 *= inputShape[i]; |
| 1955 | } |
| 1956 | |
| 1957 | TensorShape outputShape{dimension1, dimension2}; |
| 1958 | |
| 1959 | auto outInfo = ComputeReshapeInfo(outputShape, inputShape, node.output(0)); |
| 1960 | m_TensorsInfo[node.output(0)].m_info = std::make_unique<TensorInfo>(outInfo); |
| 1961 | CreateReshapeLayer(node.input(0), node.output(0), node.name()); |
| 1962 | } |
| 1963 | |
Narumol Prangnawarat | f10b15a | 2021-09-17 21:08:57 +0100 | [diff] [blame] | 1964 | void OnnxParserImpl::ParseGather(const onnx::NodeProto& node) |
| 1965 | { |
| 1966 | CHECK_VALID_SIZE(static_cast<size_t>(node.input_size()), 2); |
| 1967 | CHECK_VALID_SIZE(static_cast<size_t>(node.output_size()), 1); |
| 1968 | |
| 1969 | armnn::GatherDescriptor gatherDescriptor; |
| 1970 | gatherDescriptor.m_Axis = static_cast<int>(ReadOptionalNodeInt64Attribute(node, "axis", 0)); |
| 1971 | |
| 1972 | IConnectableLayer* layer = m_Network->AddGatherLayer(gatherDescriptor, node.name().c_str()); |
| 1973 | ARMNN_ASSERT(layer != nullptr); |
| 1974 | |
Narumol Prangnawarat | 452274c | 2021-09-23 16:12:19 +0100 | [diff] [blame] | 1975 | const TensorShape& inputShape = m_TensorsInfo[node.input(0)].m_info->GetShape(); |
| 1976 | const TensorShape& indicesShape = m_TensorsInfo[node.input(1)].m_info->GetShape(); |
| 1977 | auto outputInfo = ComputeOutputInfo({node.output(0)}, layer, { inputShape, indicesShape }, |
| 1978 | m_TensorsInfo[node.input(0)].m_dtype); |
Narumol Prangnawarat | f10b15a | 2021-09-17 21:08:57 +0100 | [diff] [blame] | 1979 | layer->GetOutputSlot(0).SetTensorInfo(outputInfo[0]); |
| 1980 | |
| 1981 | // register the input connection slots for the layer, connections are made after all layers have been created |
| 1982 | RegisterInputSlots(layer, { node.input(0), node.input(1) }); |
| 1983 | |
| 1984 | // register the output connection slots for the layer, connections are made after all layers have been created |
| 1985 | RegisterOutputSlots(layer, { node.output(0) }); |
| 1986 | } |
| 1987 | |
Narumol Prangnawarat | 1112b01 | 2021-09-30 12:10:50 +0100 | [diff] [blame] | 1988 | void OnnxParserImpl::ParseGemm(const onnx::NodeProto& node) |
| 1989 | { |
| 1990 | CHECK_VALID_SIZE(static_cast<size_t>(node.input_size()), 2, 3); |
| 1991 | CHECK_VALID_SIZE(static_cast<size_t>(node.output_size()), 1); |
| 1992 | |
| 1993 | int transA = static_cast<int>(ReadOptionalNodeUint32Attribute(node, "transA", 0)); |
| 1994 | int transB = static_cast<int>(ReadOptionalNodeUint32Attribute(node, "transB", 0)); |
| 1995 | float alpha = ReadOptionalNodeFloatAttribute(node, "alpha", 1.0); |
| 1996 | float beta = ReadOptionalNodeFloatAttribute(node, "beta", 1.0); |
| 1997 | bool biasEnabled = node.input_size() == 3; |
| 1998 | |
| 1999 | TensorShape input0Shape = m_TensorsInfo[node.input(0)].m_info->GetShape(); |
| 2000 | TensorShape input1Shape = m_TensorsInfo[node.input(1)].m_info->GetShape(); |
| 2001 | |
| 2002 | // if transB != 0, add transpose to the input1 (tanspose weight matrix in FullyConnected) |
| 2003 | armnn::FullyConnectedDescriptor fullyConnectedDescriptor; |
| 2004 | fullyConnectedDescriptor.m_BiasEnabled = biasEnabled; |
| 2005 | fullyConnectedDescriptor.m_TransposeWeightMatrix = transB; |
| 2006 | |
| 2007 | IConnectableLayer* layer = nullptr; |
| 2008 | |
| 2009 | // Just add a FullyConnected layer, weights and biases are handled as inputs now. |
| 2010 | layer = m_Network->AddFullyConnectedLayer(fullyConnectedDescriptor, node.name().c_str()); |
| 2011 | ARMNN_ASSERT(layer != nullptr); |
| 2012 | |
| 2013 | // if transA != 0, add transpose to the input0 |
| 2014 | if (transA != 0) |
| 2015 | { |
| 2016 | std::string transAName = "transpose_" + node.input(0); |
| 2017 | armnn::TransposeDescriptor transposeADescriptor; |
| 2018 | transposeADescriptor.m_DimMappings = { 1, 0 }; |
| 2019 | IConnectableLayer* transALayer = m_Network->AddTransposeLayer(transposeADescriptor, transAName.c_str()); |
| 2020 | ARMNN_ASSERT(transALayer != nullptr); |
| 2021 | auto transAInfo = ComputeOutputInfo({ transAName }, transALayer, { input0Shape }); |
| 2022 | transALayer->GetOutputSlot(0).SetTensorInfo(transAInfo[0]); |
| 2023 | transALayer->GetOutputSlot(0).Connect(layer->GetInputSlot(0u)); |
| 2024 | // register the input connection slots for the layer, connections are made after all layers have been created |
| 2025 | RegisterInputSlot(transALayer, node.input(0), 0); |
| 2026 | input0Shape = transAInfo[0].GetShape(); |
| 2027 | } |
| 2028 | else |
| 2029 | { |
| 2030 | RegisterInputSlot(layer, node.input(0), 0); |
| 2031 | } |
| 2032 | |
| 2033 | // Add constant layer to store weights/biases and connect to FullyConnected layer. |
| 2034 | if(m_TensorsInfo[node.input(1)].isConstant()) |
| 2035 | { |
| 2036 | IConnectableLayer* weightsLayer = m_Network->AddConstantLayer(CreateConstTensor(node.input(1)).first); |
| 2037 | TensorInfo weightInfo = *m_TensorsInfo[node.input(1)].m_info; |
| 2038 | weightInfo.SetConstant(); |
| 2039 | weightsLayer->GetOutputSlot(0).SetTensorInfo(weightInfo); |
| 2040 | |
| 2041 | // if alpha != 1, multiply to the weight |
| 2042 | if (alpha != 1) |
| 2043 | { |
| 2044 | std::string activationName = "activation_" + node.input(1); |
| 2045 | armnn::ActivationDescriptor activationDescriptor; |
| 2046 | activationDescriptor.m_A = alpha; |
| 2047 | activationDescriptor.m_Function = ActivationFunction::Linear; |
| 2048 | IConnectableLayer* actLayer = m_Network->AddActivationLayer(activationDescriptor, activationName.c_str()); |
| 2049 | ARMNN_ASSERT(actLayer != nullptr); |
| 2050 | |
| 2051 | auto actInfo = ComputeOutputInfo({ activationName }, actLayer, { weightInfo.GetShape() }); |
| 2052 | actLayer->GetOutputSlot(0).SetTensorInfo(actInfo[0]); |
| 2053 | actLayer->GetOutputSlot(0).Connect(layer->GetInputSlot(1u)); |
| 2054 | weightsLayer->GetOutputSlot(0).Connect(actLayer->GetInputSlot(0u)); |
| 2055 | input1Shape = actInfo[0].GetShape(); |
| 2056 | } |
| 2057 | else |
| 2058 | { |
| 2059 | weightsLayer->GetOutputSlot(0).Connect(layer->GetInputSlot(1u)); |
| 2060 | input1Shape = weightInfo.GetShape(); |
| 2061 | } |
| 2062 | } |
| 2063 | else |
| 2064 | { |
| 2065 | // if alpha != 1, multiply to the weight |
| 2066 | if (alpha != 1) |
| 2067 | { |
| 2068 | std::string activationName = "activation_" + node.input(1); |
| 2069 | armnn::ActivationDescriptor activationDescriptor; |
| 2070 | activationDescriptor.m_A = alpha; |
| 2071 | activationDescriptor.m_Function = ActivationFunction::Linear; |
| 2072 | IConnectableLayer* actLayer = m_Network->AddActivationLayer(activationDescriptor, activationName.c_str()); |
| 2073 | ARMNN_ASSERT(actLayer != nullptr); |
| 2074 | |
| 2075 | auto actInfo = ComputeOutputInfo({ activationName }, actLayer, { input1Shape }); |
| 2076 | actLayer->GetOutputSlot(0).SetTensorInfo(actInfo[0]); |
| 2077 | actLayer->GetOutputSlot(0).Connect(layer->GetInputSlot(1u)); |
| 2078 | RegisterInputSlot(actLayer, node.input(1), 0); |
| 2079 | input1Shape = actInfo[0].GetShape(); |
| 2080 | } |
| 2081 | else |
| 2082 | { |
| 2083 | RegisterInputSlot(layer, node.input(1), 1); |
| 2084 | } |
| 2085 | } |
| 2086 | |
| 2087 | if(biasEnabled && m_TensorsInfo[node.input(2)].isConstant()) |
| 2088 | { |
| 2089 | To1DTensor(node.input(2), CHECK_LOCATION()); |
| 2090 | IConnectableLayer* biasLayer = m_Network->AddConstantLayer(CreateConstTensor(node.input(2)).first); |
| 2091 | TensorInfo biasInfo = *m_TensorsInfo[node.input(2)].m_info; |
| 2092 | biasInfo.SetConstant(); |
| 2093 | biasLayer->GetOutputSlot(0).SetTensorInfo(biasInfo); |
| 2094 | |
| 2095 | // if beta != 1, multiply to the bias |
| 2096 | if (beta != 1) |
| 2097 | { |
| 2098 | std::string activationName = "activation_" + node.input(2); |
| 2099 | armnn::ActivationDescriptor activationDescriptor; |
| 2100 | activationDescriptor.m_A = beta; |
| 2101 | activationDescriptor.m_Function = ActivationFunction::Linear; |
| 2102 | IConnectableLayer* actLayer = m_Network->AddActivationLayer(activationDescriptor, activationName.c_str()); |
| 2103 | ARMNN_ASSERT(actLayer != nullptr); |
| 2104 | |
| 2105 | auto actInfo = ComputeOutputInfo({ activationName }, actLayer, { biasInfo.GetShape() }); |
| 2106 | actLayer->GetOutputSlot(0).SetTensorInfo(actInfo[0]); |
| 2107 | actLayer->GetOutputSlot(0).Connect(layer->GetInputSlot(2u)); |
| 2108 | biasLayer->GetOutputSlot(0).Connect(actLayer->GetInputSlot(0u)); |
| 2109 | } |
| 2110 | else |
| 2111 | { |
| 2112 | biasLayer->GetOutputSlot(0).Connect(layer->GetInputSlot(2u)); |
| 2113 | } |
| 2114 | } |
| 2115 | else if (biasEnabled) |
| 2116 | { |
| 2117 | // Currently we support non-constant tensor of input C (bias) of Gemm when the dimension is 1 |
| 2118 | if (m_TensorsInfo[node.input(2)].m_info->GetNumDimensions() != 1) |
| 2119 | { |
| 2120 | throw ParseException(fmt::format("The parser supports constant or non-constant with 1 dimension for " |
| 2121 | "Input C of Gemm. Input '{}' in '{}' is not supported '{}'", |
| 2122 | node.input(2), |
| 2123 | node.name(), |
| 2124 | CHECK_LOCATION().AsString())); |
| 2125 | } |
| 2126 | // if beta != 1, multiply to the bias |
| 2127 | if (beta != 1) |
| 2128 | { |
| 2129 | std::string activationName = "activation_" + node.input(2); |
| 2130 | armnn::ActivationDescriptor activationDescriptor; |
| 2131 | activationDescriptor.m_A = beta; |
| 2132 | activationDescriptor.m_Function = ActivationFunction::Linear; |
| 2133 | IConnectableLayer* actLayer = m_Network->AddActivationLayer(activationDescriptor, activationName.c_str()); |
| 2134 | ARMNN_ASSERT(actLayer != nullptr); |
| 2135 | |
| 2136 | auto actInfo = ComputeOutputInfo({ activationName }, |
| 2137 | actLayer, |
| 2138 | { m_TensorsInfo[node.input(2)].m_info->GetShape() }); |
| 2139 | actLayer->GetOutputSlot(0).SetTensorInfo(actInfo[0]); |
| 2140 | actLayer->GetOutputSlot(0).Connect(layer->GetInputSlot(2u)); |
| 2141 | RegisterInputSlot(actLayer, node.input(2), 0); |
| 2142 | } |
| 2143 | else |
| 2144 | { |
| 2145 | RegisterInputSlot(layer, node.input(2), 2); |
| 2146 | } |
| 2147 | } |
| 2148 | |
| 2149 | // Set final output of the FullyConnected layer |
| 2150 | auto outputInfo = ComputeOutputInfo({ node.output(0) }, layer, |
| 2151 | { input0Shape, input1Shape }); |
| 2152 | layer->GetOutputSlot(0).SetTensorInfo(outputInfo[0]); |
| 2153 | |
| 2154 | RegisterOutputSlots(layer, {node.output(0)}); |
| 2155 | } |
| 2156 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 2157 | void OnnxParserImpl::ParseGlobalAveragePool(const onnx::NodeProto& node) |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 2158 | { |
| 2159 | Pooling2dDescriptor desc = Pooling2dDescriptor(); |
| 2160 | desc.m_PoolType = PoolingAlgorithm::Average; |
| 2161 | |
| 2162 | //kernel size is the same as input |
| 2163 | TensorShape inputShape = m_TensorsInfo[node.input(0)].m_info->GetShape(); |
| 2164 | desc.m_PoolWidth = inputShape[3]; |
| 2165 | desc.m_PoolHeight = inputShape[2]; |
| 2166 | |
| 2167 | IConnectableLayer* layer = m_Network->AddPooling2dLayer(desc, node.name().c_str()); |
| 2168 | ARMNN_ASSERT(layer != nullptr); |
| 2169 | |
| 2170 | auto outputInfo = ComputeOutputInfo({node.output(0)}, layer, {inputShape}); |
| 2171 | layer->GetOutputSlot(0).SetTensorInfo(outputInfo[0]); |
| 2172 | |
| 2173 | // register the input connection slots for the layer, connections are made after all layers have been created |
| 2174 | // only the tensors for the inputs are relevant, exclude the const tensors |
| 2175 | RegisterInputSlots(layer, {node.input(0)}); |
| 2176 | |
| 2177 | // register the output connection slots for the layer, connections are made after all layers have been created |
| 2178 | RegisterOutputSlots(layer, {node.output(0)}); |
| 2179 | } |
| 2180 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 2181 | void OnnxParserImpl::ParseMaxPool(const onnx::NodeProto& node) |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 2182 | { |
| 2183 | Pooling2dDescriptor desc; |
| 2184 | desc.m_PoolType = PoolingAlgorithm::Max; |
| 2185 | desc.m_PaddingMethod = PaddingMethod::Exclude; |
| 2186 | AddPoolingLayer(node, desc); |
| 2187 | } |
| 2188 | |
Narumol Prangnawarat | cdc495e | 2021-09-16 18:13:39 +0100 | [diff] [blame] | 2189 | void OnnxParserImpl::ParseShape(const onnx::NodeProto& node) |
| 2190 | { |
| 2191 | CHECK_VALID_SIZE(static_cast<size_t>(node.input_size()), 1); |
| 2192 | CHECK_VALID_SIZE(static_cast<size_t>(node.output_size()), 1); |
| 2193 | |
Narumol Prangnawarat | cdc495e | 2021-09-16 18:13:39 +0100 | [diff] [blame] | 2194 | IConnectableLayer* layer = m_Network->AddShapeLayer(node.name().c_str()); |
| 2195 | ARMNN_ASSERT(layer != nullptr); |
| 2196 | |
| 2197 | TensorShape inputShape = m_TensorsInfo[node.input(0)].m_info->GetShape(); |
Narumol Prangnawarat | 452274c | 2021-09-23 16:12:19 +0100 | [diff] [blame] | 2198 | auto outputInfo = ComputeOutputInfo({node.output(0)}, layer, {inputShape}, onnx::TensorProto::INT64); |
Narumol Prangnawarat | cdc495e | 2021-09-16 18:13:39 +0100 | [diff] [blame] | 2199 | layer->GetOutputSlot(0).SetTensorInfo(outputInfo[0]); |
| 2200 | |
| 2201 | // register the input connection slots for the layer, connections are made after all layers have been created |
| 2202 | RegisterInputSlots(layer, {node.input(0)}); |
| 2203 | |
| 2204 | // register the output connection slots for the layer, connections are made after all layers have been created |
| 2205 | RegisterOutputSlots(layer, {node.output(0)}); |
| 2206 | } |
| 2207 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 2208 | void OnnxParserImpl::ParseReshape(const onnx::NodeProto& node) |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 2209 | { |
| 2210 | CHECK_VALID_SIZE(static_cast<size_t>(node.input_size()), 2); |
| 2211 | CHECK_VALID_SIZE(static_cast<size_t>(node.output_size()), 1); |
| 2212 | |
| 2213 | CHECK_VALID_DATATYPE(node.name(), node.input(0), |
| 2214 | m_TensorsInfo[node.input(0)].m_dtype, |
| 2215 | onnx::TensorProto::FLOAT); //input |
| 2216 | CHECK_VALID_DATATYPE(node.name(), node.input(1), |
| 2217 | m_TensorsInfo[node.input(1)].m_dtype, |
| 2218 | onnx::TensorProto::INT64); //shape |
| 2219 | |
Narumol Prangnawarat | 4b536e3 | 2021-10-18 12:35:19 +0100 | [diff] [blame] | 2220 | TensorShape inputShape = m_TensorsInfo[node.input(0)].m_info->GetShape(); |
| 2221 | |
| 2222 | std::vector<unsigned int> targetShape; |
| 2223 | if(m_TensorsInfo[node.input(1)].isConstant()) |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 2224 | { |
Narumol Prangnawarat | 4b536e3 | 2021-10-18 12:35:19 +0100 | [diff] [blame] | 2225 | unsigned int dims = static_cast<unsigned int>(m_TensorsInfo[node.input(1)].m_tensor->int64_data_size()); |
| 2226 | targetShape.reserve(dims); |
| 2227 | |
| 2228 | for(uint i = 0; i < dims; i++) |
| 2229 | { |
| 2230 | int val = CHECKED_INT32(m_TensorsInfo[node.input(1)].m_tensor->int64_data(static_cast<int>(i))); |
| 2231 | targetShape[i]= static_cast<unsigned int>(val); |
| 2232 | } |
| 2233 | } |
| 2234 | else |
| 2235 | { |
| 2236 | // The parser only supports shape (batch, -1) or (-1) for non-constant shape input. |
| 2237 | unsigned int dims = m_TensorsInfo[node.input(1)].m_info->GetNumDimensions(); |
| 2238 | TensorShape shapes = m_TensorsInfo[node.input(1)].m_info->GetShape(); |
| 2239 | if (dims != 1 || shapes[0] > 2) |
| 2240 | { |
| 2241 | throw ParseException(fmt::format("Invalid input shape '{}' in Reshape layer '{}' {}", |
| 2242 | node.input(1), |
| 2243 | node.name(), |
| 2244 | CHECK_LOCATION().AsString())); |
| 2245 | } |
| 2246 | |
| 2247 | unsigned int numInputElements = m_TensorsInfo[node.input(0)].m_info->GetNumElements(); |
| 2248 | if (shapes[0] == 1) |
| 2249 | { |
| 2250 | targetShape = { numInputElements }; |
| 2251 | } |
| 2252 | else if (shapes[0] == 2) |
| 2253 | { |
| 2254 | targetShape = { inputShape[0] , numInputElements / inputShape[0] }; |
| 2255 | } |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 2256 | } |
| 2257 | |
| 2258 | if(m_TensorsInfo[node.input(0)].isConstant()) |
| 2259 | { |
| 2260 | //make a new cst tensor -> move the data to the output tensor (the shape is already good in the output tensor) |
| 2261 | if(m_TensorsInfo.count(node.output(0)) == 0) |
| 2262 | { |
| 2263 | m_TensorsInfo[node.output(0)] = OnnxTensor(); |
| 2264 | } |
| 2265 | m_TensorsInfo[node.output(0)].m_tensor = |
| 2266 | std::make_unique<onnx::TensorProto>(*m_TensorsInfo[node.input(0)].m_tensor); |
| 2267 | } |
| 2268 | else |
| 2269 | { |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 2270 | if(m_TensorsInfo.count(node.output(0)) == 0 || m_TensorsInfo[node.output(0)].m_info == nullptr) |
| 2271 | { |
Narumol Prangnawarat | 4b536e3 | 2021-10-18 12:35:19 +0100 | [diff] [blame] | 2272 | auto outInfo = ComputeReshapeInfo( |
| 2273 | TensorShape(static_cast<unsigned int>(targetShape.size()), targetShape.data()), |
| 2274 | inputShape, node.output(0)); |
Ryan OShea | ed27ee7 | 2020-04-22 16:37:29 +0100 | [diff] [blame] | 2275 | m_TensorsInfo[node.output(0)].m_info = std::make_unique<TensorInfo>(outInfo); |
| 2276 | } |
| 2277 | |
| 2278 | CreateReshapeLayer(node.input(0), node.output(0), node.name()); |
| 2279 | } |
| 2280 | } |
| 2281 | |
Narumol Prangnawarat | fe6aa2f | 2021-09-23 16:11:17 +0100 | [diff] [blame] | 2282 | void OnnxParserImpl::ParseUnsqueeze(const onnx::NodeProto& node) |
| 2283 | { |
| 2284 | CHECK_VALID_SIZE(armnn::numeric_cast<size_t>(node.input_size()), 1, 2); |
| 2285 | CHECK_VALID_SIZE(armnn::numeric_cast<size_t>(node.output_size()), 1); |
| 2286 | |
Narumol Prangnawarat | fe6aa2f | 2021-09-23 16:11:17 +0100 | [diff] [blame] | 2287 | TensorShape inputShape = m_TensorsInfo[node.input(0)].m_info->GetShape(); |
| 2288 | std::vector<uint32_t> dims; |
| 2289 | if (node.input_size() == 1 && node.attribute_size() > 0) |
| 2290 | { |
| 2291 | dims = ReadMandatoryNodeUint32ListAttribute(node, "axes"); |
| 2292 | } |
| 2293 | else |
| 2294 | { |
| 2295 | CHECK_VALID_DATATYPE(node.name(), node.input(1), |
| 2296 | m_TensorsInfo[node.input(1)].m_dtype, |
| 2297 | onnx::TensorProto::INT64); //axes |
| 2298 | |
| 2299 | auto int64Axes = m_TensorsInfo[node.input(1)].m_tensor->int64_data().data(); |
| 2300 | uint numDim = armnn::numeric_cast<uint>(m_TensorsInfo[node.input(1)].m_tensor->int64_data_size()); |
| 2301 | |
| 2302 | for(uint i = 0; i < numDim; i++) |
| 2303 | { |
| 2304 | uint32_t uint32Value = CHECKED_NON_NEGATIVE(CHECKED_INT32(int64Axes[i])); |
| 2305 | dims.push_back(uint32Value); |
| 2306 | } |
| 2307 | } |
| 2308 | |
| 2309 | // Ensure that the axes are sorted |
| 2310 | std::sort(dims.begin(), dims.end()); |
| 2311 | |
| 2312 | std::vector<unsigned int> targetShape; |
| 2313 | |
Narumol Prangnawarat | 452274c | 2021-09-23 16:12:19 +0100 | [diff] [blame] | 2314 | if (inputShape.GetDimensionality() != Dimensionality::Scalar) |
Narumol Prangnawarat | fe6aa2f | 2021-09-23 16:11:17 +0100 | [diff] [blame] | 2315 | { |
Narumol Prangnawarat | 452274c | 2021-09-23 16:12:19 +0100 | [diff] [blame] | 2316 | for(uint i = 0; i < inputShape.GetNumDimensions(); i++) |
| 2317 | { |
| 2318 | targetShape.push_back(inputShape[i]); |
| 2319 | } |
Narumol Prangnawarat | fe6aa2f | 2021-09-23 16:11:17 +0100 | [diff] [blame] | 2320 | } |
| 2321 | |
| 2322 | for(uint i = 0; i < dims.size(); i++) |
| 2323 | { |
| 2324 | targetShape.insert(targetShape.begin() + armnn::numeric_cast<int>(dims[i]), 1); |
| 2325 | } |
| 2326 | |
Narumol Prangnawarat | 452274c | 2021-09-23 16:12:19 +0100 | [diff] [blame] | 2327 | auto outInfo = ComputeReshapeInfo(TensorShape(static_cast<unsigned int>(targetShape.size()), targetShape.data()), |
| 2328 | inputShape, node.output(0), m_TensorsInfo[node.input(0)].m_info->GetDataType()); |
Narumol Prangnawarat | fe6aa2f | 2021-09-23 16:11:17 +0100 | [diff] [blame] | 2329 | m_TensorsInfo[node.output(0)].m_info = std::make_unique<TensorInfo>(outInfo); |
Narumol Prangnawarat | 452274c | 2021-09-23 16:12:19 +0100 | [diff] [blame] | 2330 | m_TensorsInfo[node.output(0)].m_dtype = m_TensorsInfo[node.input(0)].m_dtype; |
Narumol Prangnawarat | fe6aa2f | 2021-09-23 16:11:17 +0100 | [diff] [blame] | 2331 | |
| 2332 | CreateReshapeLayer(node.input(0), node.output(0), node.name()); |
| 2333 | } |
| 2334 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 2335 | void OnnxParserImpl::PrependForBroadcast(const std::string& outputName, |
| 2336 | const std::string& input0, |
| 2337 | const std::string& input1) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2338 | { |
| 2339 | //input0 should be reshaped to have same number of dim as input1 |
| 2340 | TensorInfo outputTensorInfo = TensorInfo(*m_TensorsInfo[input0].m_info); |
| 2341 | |
| 2342 | TensorShape input0Shape = m_TensorsInfo[input0].m_info->GetShape(); |
| 2343 | TensorShape input1Shape = m_TensorsInfo[input1].m_info->GetShape(); |
| 2344 | |
| 2345 | uint32_t diff = input1Shape.GetNumDimensions() - input0Shape.GetNumDimensions(); |
| 2346 | std::vector<uint32_t> newShape; |
| 2347 | while(diff > 0) |
| 2348 | { |
| 2349 | newShape.push_back(1); |
| 2350 | diff--; |
| 2351 | } |
| 2352 | for (uint dim = 0; dim < input0Shape.GetNumDimensions(); ++dim) |
| 2353 | { |
| 2354 | newShape.push_back(input0Shape[dim]); |
| 2355 | } |
| 2356 | outputTensorInfo.SetShape(TensorShape(static_cast<unsigned int>(newShape.size()), newShape.data())); |
| 2357 | |
| 2358 | //add the new tensor to m_TensorsInfo |
| 2359 | m_TensorsInfo[outputName] = OnnxTensor(); |
| 2360 | m_TensorsInfo[outputName].m_info = std::make_unique<TensorInfo>(outputTensorInfo); |
| 2361 | |
| 2362 | //add reshape layer if the parent was not constant... |
| 2363 | if( ! m_TensorsInfo[input0].isConstant()) |
| 2364 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 2365 | CreateReshapeLayer(input0, outputName, fmt::format("Add:reshapeOf{}", input0)); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2366 | } |
| 2367 | else //make it constant and it will be create in Add |
| 2368 | { |
| 2369 | m_TensorsInfo[outputName].m_tensor = std::make_unique<onnx::TensorProto>(*m_TensorsInfo[input0].m_tensor); |
| 2370 | |
| 2371 | } |
| 2372 | } |
| 2373 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 2374 | void OnnxParserImpl::SetupInputLayers() |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2375 | { |
| 2376 | //Find user input and add their layers |
| 2377 | for(int inputIndex = 0; inputIndex < m_Graph->input_size(); ++inputIndex) |
| 2378 | { |
| 2379 | auto input = m_Graph->input(inputIndex); |
Narumol Prangnawarat | 1b11f32 | 2021-10-13 11:44:50 +0100 | [diff] [blame] | 2380 | if (!m_TensorsInfo[input.name()].isConstant()) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2381 | { |
| 2382 | IConnectableLayer* layer = |
Narumol Prangnawarat | 1b11f32 | 2021-10-13 11:44:50 +0100 | [diff] [blame] | 2383 | m_Network->AddInputLayer(static_cast<armnn::LayerBindingId>(inputIndex), input.name().c_str()); |
| 2384 | TensorInfo tensorInfo = *m_TensorsInfo[input.name()].m_info; |
| 2385 | if (tensorInfo.GetShape().GetDimensionality() == Dimensionality::NotSpecified) |
| 2386 | { |
| 2387 | if (m_InputShapes.find(input.name()) == m_InputShapes.end()) |
| 2388 | { |
| 2389 | throw ParseException(fmt::format("The parser does not support dynamic tensor, " |
| 2390 | "please specify input shape for {}. {}", |
| 2391 | input.name(), |
| 2392 | CHECK_LOCATION().AsString())); |
| 2393 | } |
| 2394 | else |
| 2395 | { |
| 2396 | tensorInfo.SetShape(m_InputShapes[input.name()]); |
| 2397 | m_TensorsInfo[input.name()].m_info = std::make_unique<TensorInfo>(tensorInfo); |
| 2398 | } |
| 2399 | |
| 2400 | } |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2401 | layer->GetOutputSlot(0).SetTensorInfo(tensorInfo); |
| 2402 | |
Narumol Prangnawarat | 1b11f32 | 2021-10-13 11:44:50 +0100 | [diff] [blame] | 2403 | m_InputInfos[input.name()] = tensorInfo; |
| 2404 | |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2405 | RegisterOutputSlots(layer,{ input.name() }); |
| 2406 | } |
| 2407 | } |
| 2408 | } |
| 2409 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 2410 | void OnnxParserImpl::SetupOutputLayers() |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2411 | { |
| 2412 | if(m_Graph->output_size() == 0) |
| 2413 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 2414 | throw ParseException(fmt::format("The given model does not have any outputs {}", CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2415 | } |
| 2416 | |
| 2417 | for(int outputIndex = 0; outputIndex < m_Graph->output_size(); ++outputIndex) |
| 2418 | { |
| 2419 | IConnectableLayer* layer = |
| 2420 | m_Network->AddOutputLayer(static_cast<armnn::LayerBindingId>(outputIndex), |
| 2421 | m_Graph->output(outputIndex).name().c_str()); |
| 2422 | |
| 2423 | RegisterInputSlots(layer, { m_Graph->output(outputIndex).name() }); |
| 2424 | } |
| 2425 | } |
| 2426 | |
Narumol Prangnawarat | 1112b01 | 2021-09-30 12:10:50 +0100 | [diff] [blame] | 2427 | void OnnxParserImpl::RegisterInputSlot(IConnectableLayer* layer, |
| 2428 | const std::string& tensorId, |
| 2429 | unsigned int slotIndex) |
| 2430 | { |
| 2431 | armnn::IInputSlot* slot = &(layer->GetInputSlot(slotIndex)); |
| 2432 | |
| 2433 | auto it = m_TensorConnections.find(tensorId); |
| 2434 | |
| 2435 | if (it == m_TensorConnections.end()) |
| 2436 | { |
Narumol Prangnawarat | 1b11f32 | 2021-10-13 11:44:50 +0100 | [diff] [blame] | 2437 | //First time seeing this tensor, we need to map it |
Narumol Prangnawarat | 1112b01 | 2021-09-30 12:10:50 +0100 | [diff] [blame] | 2438 | m_TensorConnections[tensorId] = TensorSlots(); |
| 2439 | } |
| 2440 | m_TensorConnections[tensorId].inputSlots.push_back(slot); |
| 2441 | } |
| 2442 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 2443 | void OnnxParserImpl::RegisterInputSlots(IConnectableLayer* layer, const std::vector<std::string>& tensorIds) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2444 | { |
Narumol Prangnawarat | ac2770a | 2020-04-01 16:51:23 +0100 | [diff] [blame] | 2445 | ARMNN_ASSERT(layer != nullptr); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2446 | if (tensorIds.size() != layer->GetNumInputSlots()) |
| 2447 | { |
| 2448 | throw ParseException( |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 2449 | fmt::format("The number of tensor inputs ({}) does not match the number expected ({}) {}", |
| 2450 | tensorIds.size(), |
| 2451 | layer->GetNumInputSlots(), |
| 2452 | CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2453 | } |
Matthew Sloyan | 81beae3 | 2021-07-13 19:46:11 +0100 | [diff] [blame] | 2454 | |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2455 | for (unsigned int slotIndex = 0; slotIndex < layer->GetNumInputSlots(); ++slotIndex) |
| 2456 | { |
| 2457 | std::string tensorId = tensorIds[slotIndex]; |
| 2458 | armnn::IInputSlot* slot = &(layer->GetInputSlot(slotIndex)); |
| 2459 | |
| 2460 | auto it = m_TensorConnections.find(tensorId); |
| 2461 | |
| 2462 | if (it == m_TensorConnections.end()) |
| 2463 | { |
Narumol Prangnawarat | 1b11f32 | 2021-10-13 11:44:50 +0100 | [diff] [blame] | 2464 | // First time seing this tensor, we need to map it |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2465 | m_TensorConnections[tensorId] = TensorSlots(); |
| 2466 | } |
| 2467 | m_TensorConnections[tensorId].inputSlots.push_back(slot); |
| 2468 | } |
| 2469 | } |
| 2470 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 2471 | void OnnxParserImpl::RegisterOutputSlots(IConnectableLayer* layer, const std::vector<std::string>& tensorIds) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2472 | { |
Narumol Prangnawarat | ac2770a | 2020-04-01 16:51:23 +0100 | [diff] [blame] | 2473 | ARMNN_ASSERT(layer != nullptr); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2474 | if (tensorIds.size() != layer->GetNumOutputSlots()) |
| 2475 | { |
| 2476 | throw ParseException( |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 2477 | fmt::format("The number of tensor outputs ({}) does not match the number expected ({}) {} ", |
| 2478 | tensorIds.size(), |
| 2479 | layer->GetNumOutputSlots(), |
| 2480 | CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2481 | } |
| 2482 | |
| 2483 | for (unsigned int slotIndex = 0; slotIndex < layer->GetNumOutputSlots(); ++slotIndex) |
| 2484 | { |
| 2485 | std::string tensorId = tensorIds[slotIndex]; |
| 2486 | armnn::IOutputSlot* slot = &(layer->GetOutputSlot(slotIndex)); |
| 2487 | |
| 2488 | auto it = m_TensorConnections.find(tensorId); |
| 2489 | |
| 2490 | if (it == m_TensorConnections.end()) |
| 2491 | { |
| 2492 | //First time seing this tensor, we need to map it |
| 2493 | m_TensorConnections[tensorId] = TensorSlots(); |
| 2494 | } |
| 2495 | |
Ryan OShea | 337c17f | 2020-02-21 12:33:17 +0000 | [diff] [blame] | 2496 | TensorSlots& tensorSlots = m_TensorConnections[tensorId]; |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2497 | |
| 2498 | // assuming there is only one producer for that tensor |
| 2499 | if (tensorSlots.outputSlot != nullptr) |
| 2500 | { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 2501 | throw ParseException(fmt::format("Another layer has already registered itself as the producer of " |
| 2502 | "tensor:{} {}", |
| 2503 | tensorId, |
| 2504 | CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2505 | } |
| 2506 | tensorSlots.outputSlot = slot; |
| 2507 | } |
Narumol Prangnawarat | 1b11f32 | 2021-10-13 11:44:50 +0100 | [diff] [blame] | 2508 | |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2509 | } |
| 2510 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 2511 | BindingPointInfo OnnxParserImpl::GetNetworkInputBindingInfo(const std::string& name) const |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2512 | { |
| 2513 | for(int i = 0; i < m_Graph->input_size(); ++i) |
| 2514 | { |
| 2515 | auto input = m_Graph->input(i); |
| 2516 | if(input.name() == name) |
| 2517 | { |
Narumol Prangnawarat | 1b11f32 | 2021-10-13 11:44:50 +0100 | [diff] [blame] | 2518 | auto it = m_InputInfos.find(name); |
| 2519 | |
| 2520 | if (it != m_InputInfos.end()) |
| 2521 | { |
| 2522 | return std::make_pair(static_cast<armnn::LayerBindingId>(i), it->second); |
| 2523 | } |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2524 | } |
| 2525 | } |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 2526 | throw InvalidArgumentException(fmt::format("The input layer '{}' does not exist {}", |
| 2527 | name, CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2528 | } |
| 2529 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 2530 | BindingPointInfo OnnxParserImpl::GetNetworkOutputBindingInfo(const std::string& name) const |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2531 | { |
| 2532 | for(int i = 0; i < m_Graph->output_size(); ++i) |
| 2533 | { |
| 2534 | auto output = m_Graph->output(i); |
| 2535 | if(output.name() == name) |
| 2536 | { |
Narumol Prangnawarat | 1b11f32 | 2021-10-13 11:44:50 +0100 | [diff] [blame] | 2537 | auto it = m_OutputInfos.find(name); |
| 2538 | |
| 2539 | if (it != m_OutputInfos.end()) |
| 2540 | { |
| 2541 | return std::make_pair(static_cast<armnn::LayerBindingId>(i), it->second); |
| 2542 | } |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2543 | } |
| 2544 | } |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 2545 | throw InvalidArgumentException(fmt::format("The output layer '{}' does not exist {}", |
| 2546 | name, CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2547 | } |
| 2548 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 2549 | std::vector<std::string> OnnxParserImpl::GetInputs(ModelPtr& model) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2550 | { |
| 2551 | if(model == nullptr) { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 2552 | throw InvalidArgumentException(fmt::format("The given model cannot be null {}", |
| 2553 | CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2554 | } |
| 2555 | |
| 2556 | std::vector<std::string> inputNames; |
| 2557 | std::map<std::string, bool> isConstant; |
| 2558 | for(auto tensor : model->graph().initializer()) |
| 2559 | { |
| 2560 | isConstant[tensor.name()] = true; |
| 2561 | } |
| 2562 | for(auto input : model->graph().input()) |
| 2563 | { |
| 2564 | auto it = isConstant.find(input.name()); |
| 2565 | if(it == isConstant.end()) |
| 2566 | { |
| 2567 | inputNames.push_back(input.name()); |
| 2568 | } |
| 2569 | } |
| 2570 | return inputNames; |
| 2571 | } |
| 2572 | |
Kevin May | ef33cb1 | 2021-01-29 14:24:57 +0000 | [diff] [blame] | 2573 | std::vector<std::string> OnnxParserImpl::GetOutputs(ModelPtr& model) |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2574 | { |
| 2575 | if(model == nullptr) { |
James Ward | 58dec6b | 2020-09-11 17:32:44 +0100 | [diff] [blame] | 2576 | throw InvalidArgumentException(fmt::format("The given model cannot be null {}", |
| 2577 | CHECK_LOCATION().AsString())); |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2578 | } |
| 2579 | |
| 2580 | std::vector<std::string> outputNames; |
| 2581 | for(auto output : model->graph().output()) |
| 2582 | { |
| 2583 | outputNames.push_back(output.name()); |
| 2584 | } |
| 2585 | return outputNames; |
| 2586 | } |
| 2587 | |
Matthew Sloyan | ac001ee | 2021-02-03 10:43:04 +0000 | [diff] [blame] | 2588 | const std::string OnnxParserImpl::GetVersion() |
| 2589 | { |
| 2590 | return ONNX_PARSER_VERSION; |
| 2591 | } |
| 2592 | |
telsoa01 | c577f2c | 2018-08-31 09:22:23 +0100 | [diff] [blame] | 2593 | } // namespace armnnOnnxParser |