Gitiles
Code Review Sign In
review.mlplatform.org / ml / ComputeLibrary / f07d28d9ee8ae73a93fe433f72855b6dcf58ad90 / . / examples / graph_inception_v3.cpp
blob: 88a0325b639ee2496f7dec125fe0ab74e5f87542 [file] [log] [blame]
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]1/*
2 * Copyright (c) 2017-2018 ARM Limited.
3 *
4 * SPDX-License-Identifier: MIT
5 *
6 * Permission is hereby granted, free of charge, to any person obtaining a copy
7 * of this software and associated documentation files (the "Software"), to
8 * deal in the Software without restriction, including without limitation the
9 * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
10 * sell copies of the Software, and to permit persons to whom the Software is
11 * furnished to do so, subject to the following conditions:
12 *
13 * The above copyright notice and this permission notice shall be included in all
14 * copies or substantial portions of the Software.
15 *
16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
19 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
21 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
22 * SOFTWARE.
23 */
24#include "arm_compute/graph/Graph.h"
25#include "arm_compute/graph/Nodes.h"
26#include "arm_compute/graph/SubGraph.h"
27#include "support/ToolchainSupport.h"
28#include "utils/GraphUtils.h"
29#include "utils/Utils.h"
30
31#include <cstdlib>
32#include <tuple>
33
34using namespace arm_compute::utils;
35using namespace arm_compute::graph;
36using namespace arm_compute::graph_utils;
37
38/** Example demonstrating how to implement InceptionV3's network using the Compute Library's graph API
39 *
40 * @param[in] argc Number of arguments
41 * @param[in] argv Arguments ( [optional] Path to the weights folder, [optional] image, [optional] labels )
42 */
43class InceptionV3Example : public Example
44{
45public:
46 void do_setup(int argc, char **argv) override
47 {
48 std::string data_path; /* Path to the trainable data */
49 std::string image; /* Image data */
50 std::string label; /* Label data */
51
52 constexpr float mean = 0.f; /* Mean value to subtract from the channels */
53 constexpr float std = 255.f; /* Standard deviation value to divide from the channels */
54
55 // Set target. 0 (NEON), 1 (OpenCL). By default it is NEON
Gian Marco2d405552018-02-05 08:54:54 +0000[diff] [blame]56 TargetHint target_hint = set_target_hint(argc > 1 ? std::strtol(argv[1], nullptr, 10) : 0);
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]57
58 // Parse arguments
59 if(argc < 2)
60 {
61 // Print help
62 std::cout << "Usage: " << argv[0] << " [target] [path_to_data] [image] [labels]\n\n";
63 std::cout << "No data folder provided: using random values\n\n";
64 }
65 else if(argc == 2)
66 {
67 std::cout << "Usage: " << argv[0] << " " << argv[1] << " [path_to_data] [image] [labels]\n\n";
68 std::cout << "No data folder provided: using random values\n\n";
69 }
70 else if(argc == 3)
71 {
72 data_path = argv[2];
73 std::cout << "Usage: " << argv[0] << " " << argv[1] << " " << argv[2] << " [image] [labels]\n\n";
74 std::cout << "No image provided: using random values\n\n";
75 }
76 else if(argc == 4)
77 {
78 data_path = argv[2];
79 image = argv[3];
80 std::cout << "Usage: " << argv[0] << " " << argv[1] << " " << argv[2] << " " << argv[3] << " [labels]\n\n";
81 std::cout << "No text file with labels provided: skipping output accessor\n\n";
82 }
83 else
84 {
85 data_path = argv[2];
86 image = argv[3];
87 label = argv[4];
88 }
89
Gian Marco2d405552018-02-05 08:54:54 +0000[diff] [blame]90 graph << target_hint << Tensor(TensorInfo(TensorShape(299U, 299U, 3U, 1U), 1, DataType::F32),
91 get_input_accessor(image,
92 mean, mean, mean,
93 std, std, std, false /* Do not convert to BGR */))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]94
95 << ConvolutionLayer(3U, 3U, 32U,
96 get_weights_accessor(data_path, "/cnn_data/inceptionv3_model/Conv2d_1a_3x3_weights.npy"),
97 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr), PadStrideInfo(2, 2, 0, 0))
98 << BatchNormalizationLayer(get_weights_accessor(data_path,
99 "/cnn_data/inceptionv3_model/Conv2d_1a_3x3_BatchNorm_moving_mean.npy"),
100 get_weights_accessor(data_path,
101 "/cnn_data/inceptionv3_model/Conv2d_1a_3x3_BatchNorm_moving_variance.npy"),
102 get_random_accessor(1.f, 1.f), get_weights_accessor(data_path,
103 "/cnn_data/inceptionv3_model/Conv2d_1a_3x3_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]104 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]105
106 << ConvolutionLayer(3U, 3U, 32U,
107 get_weights_accessor(data_path, "/cnn_data/inceptionv3_model/Conv2d_2a_3x3_weights.npy"),
108 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr), PadStrideInfo(1, 1, 0, 0))
109 << BatchNormalizationLayer(get_weights_accessor(data_path,
110 "/cnn_data/inceptionv3_model/Conv2d_2a_3x3_BatchNorm_moving_mean.npy"),
111 get_weights_accessor(data_path,
112 "/cnn_data/inceptionv3_model/Conv2d_2a_3x3_BatchNorm_moving_variance.npy"),
113 get_random_accessor(1.f, 1.f), get_weights_accessor(data_path,
114 "/cnn_data/inceptionv3_model/Conv2d_2a_3x3_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]115 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]116
117 << ConvolutionLayer(3U, 3U, 64U,
118 get_weights_accessor(data_path, "/cnn_data/inceptionv3_model/Conv2d_2b_3x3_weights.npy"),
119 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr), PadStrideInfo(1, 1, 1, 1))
120 << BatchNormalizationLayer(get_weights_accessor(data_path,
121 "/cnn_data/inceptionv3_model/Conv2d_2b_3x3_BatchNorm_moving_mean.npy"),
122 get_weights_accessor(data_path,
123 "/cnn_data/inceptionv3_model/Conv2d_2b_3x3_BatchNorm_moving_variance.npy"),
124 get_random_accessor(1.f, 1.f), get_weights_accessor(data_path,
125 "/cnn_data/inceptionv3_model/Conv2d_2b_3x3_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]126 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]127
128 << PoolingLayer(PoolingLayerInfo(PoolingType::MAX, 3, PadStrideInfo(2, 2, 0, 0, DimensionRoundingType::CEIL)))
129
130 << ConvolutionLayer(1U, 1U, 80U,
131 get_weights_accessor(data_path, "/cnn_data/inceptionv3_model/Conv2d_3b_1x1_weights.npy"),
132 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr), PadStrideInfo(1, 1, 0, 0))
133 << BatchNormalizationLayer(get_weights_accessor(data_path,
134 "/cnn_data/inceptionv3_model/Conv2d_3b_1x1_BatchNorm_moving_mean.npy"),
135 get_weights_accessor(data_path,
136 "/cnn_data/inceptionv3_model/Conv2d_3b_1x1_BatchNorm_moving_variance.npy"),
137 get_random_accessor(1.f, 1.f), get_weights_accessor(data_path,
138 "/cnn_data/inceptionv3_model/Conv2d_3b_1x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]139 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]140
141 << ConvolutionLayer(3U, 3U, 192U,
142 get_weights_accessor(data_path, "/cnn_data/inceptionv3_model/Conv2d_4a_3x3_weights.npy"),
143 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr), PadStrideInfo(1, 1, 0, 0))
144 << BatchNormalizationLayer(get_weights_accessor(data_path,
145 "/cnn_data/inceptionv3_model/Conv2d_4a_3x3_BatchNorm_moving_mean.npy"),
146 get_weights_accessor(data_path,
147 "/cnn_data/inceptionv3_model/Conv2d_4a_3x3_BatchNorm_moving_variance.npy"),
148 get_random_accessor(1.f, 1.f), get_weights_accessor(data_path,
149 "/cnn_data/inceptionv3_model/Conv2d_4a_3x3_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]150 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]151
152 << PoolingLayer(PoolingLayerInfo(PoolingType::MAX, 3, PadStrideInfo(2, 2, 0, 0, DimensionRoundingType::CEIL)))
153
154 << get_inception_node_A(data_path, "Mixed_5b", 64U, std::make_tuple(48U, 64U), std::make_tuple(64U, 96U, 96U),
155 32U)
156 << get_inception_node_A(data_path, "Mixed_5c", 64U, std::make_tuple(48U, 64U), std::make_tuple(64U, 96U, 96U),
157 64U, true)
158 << get_inception_node_A(data_path, "Mixed_5d", 64U, std::make_tuple(48U, 64U), std::make_tuple(64U, 96U, 96U),
159 64U)
160
161 << get_inception_node_B(data_path, "Mixed_6a", 384U, std::make_tuple(64U, 96U, 96U))
162
163 << get_inception_node_C(data_path, "Mixed_6b", 192U, std::make_tuple(128U, 128U, 192U),
164 std::make_tuple(128U, 128U, 128U, 128U, 192U), 192U)
165 << get_inception_node_C(data_path, "Mixed_6c", 192U, std::make_tuple(160U, 160U, 192U),
166 std::make_tuple(160U, 160U, 160U, 160U, 192U), 192U)
167 << get_inception_node_C(data_path, "Mixed_6d", 192U, std::make_tuple(160U, 160U, 192U),
168 std::make_tuple(160U, 160U, 160U, 160U, 192U), 192U)
169 << get_inception_node_C(data_path, "Mixed_6e", 192U, std::make_tuple(192U, 192U, 192U),
170 std::make_tuple(192U, 192U, 192U, 192U, 192U), 192U)
171
172 << get_inception_node_D(data_path, "Mixed_7a", std::make_tuple(192U, 320U),
173 std::make_tuple(192U, 192U, 192U, 192U))
174
175 << get_inception_node_E(data_path, "Mixed_7b", 320U, std::make_tuple(384U, 384U, 384U),
176 std::make_tuple(448U, 384U, 384U, 384U), 192U)
177 << get_inception_node_E(data_path, "Mixed_7c", 320U, std::make_tuple(384U, 384U, 384U),
178 std::make_tuple(448U, 384U, 384U, 384U), 192U, true)
179
180 << PoolingLayer(PoolingLayerInfo(PoolingType::AVG, 8, PadStrideInfo(1, 1, 0, 0, DimensionRoundingType::CEIL)))
181 << ConvolutionLayer(1U, 1U, 1001U, get_weights_accessor(data_path,
182 "/cnn_data/inceptionv3_model/Logits_Conv2d_1c_1x1_weights.npy"),
183 get_weights_accessor(data_path,
184 "/cnn_data/inceptionv3_model/Logits_Conv2d_1c_1x1_biases.npy"),
185 PadStrideInfo(1, 1, 0, 0))
186 << ReshapeLayer(TensorShape(1001U)) << SoftmaxLayer()
187 << Tensor(get_output_accessor(label, 5));
188 }
189
190 void do_run() override
191 {
192 graph.run();
193 }
194
195private:
196 Graph graph{};
197
198private:
199 BranchLayer get_inception_node_A(const std::string &data_path, std::string &&param_path,
200 unsigned int a_filt,
201 std::tuple<unsigned int, unsigned int> b_filters,
202 std::tuple<unsigned int, unsigned int, unsigned int> c_filters,
203 unsigned int d_filt,
204 bool is_name_different = false)
205 {
206 std::string total_path = "/cnn_data/inceptionv3_model/" + param_path + "_";
207 std::cout << total_path << std::endl;
208
209 // This is due to a naming issue in the tf model
210 std::string conv_id0 = "_0a_";
211 std::string conv_id1 = "2d_0b_";
212 if(is_name_different)
213 {
214 conv_id0 = "_0b_";
215 conv_id1 = "_1_0c_";
216 }
217
218 SubGraph i_a;
219 i_a << ConvolutionLayer(
220 1U, 1U, a_filt,
221 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_0a_1x1_weights.npy"),
222 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
223 PadStrideInfo(1, 1, 0, 0))
224 << BatchNormalizationLayer(
225 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_0a_1x1_BatchNorm_moving_mean.npy"),
226 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_0a_1x1_BatchNorm_moving_variance.npy"),
227 get_random_accessor(1.f, 1.f),
228 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_0a_1x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]229 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU));
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]230
231 SubGraph i_b;
232 i_b << ConvolutionLayer(
233 1U, 1U, std::get<0>(b_filters),
234 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d" + conv_id0 + "1x1_weights.npy"),
235 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
236 PadStrideInfo(1, 1, 0, 0))
237 << BatchNormalizationLayer(
238 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d" + conv_id0 + "1x1_BatchNorm_moving_mean.npy"),
239 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d" + conv_id0 + "1x1_BatchNorm_moving_variance.npy"),
240 get_random_accessor(1.f, 1.f),
241 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d" + conv_id0 + "1x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]242 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]243 << ConvolutionLayer(
244 5U, 5U, std::get<1>(b_filters),
245 get_weights_accessor(data_path, total_path + "Branch_1_Conv" + conv_id1 + "5x5_weights.npy"),
246 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
247 PadStrideInfo(1, 1, 2, 2))
248 << BatchNormalizationLayer(
249 get_weights_accessor(data_path, total_path + "Branch_1_Conv" + conv_id1 + "5x5_BatchNorm_moving_mean.npy"),
250 get_weights_accessor(data_path, total_path + "Branch_1_Conv" + conv_id1 + "5x5_BatchNorm_moving_variance.npy"),
251 get_random_accessor(1.f, 1.f),
252 get_weights_accessor(data_path, total_path + "Branch_1_Conv" + conv_id1 + "5x5_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]253 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU));
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]254
255 SubGraph i_c;
256 i_c << ConvolutionLayer(
257 1U, 1U, std::get<0>(c_filters),
258 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0a_1x1_weights.npy"),
259 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
260 PadStrideInfo(1, 1, 0, 0))
261 << BatchNormalizationLayer(
262 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0a_1x1_BatchNorm_moving_mean.npy"),
263 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0a_1x1_BatchNorm_moving_variance.npy"),
264 get_random_accessor(1.f, 1.f),
265 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0a_1x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]266 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]267 << ConvolutionLayer(
268 3U, 3U, std::get<1>(c_filters),
269 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0b_3x3_weights.npy"),
270 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
271 PadStrideInfo(1, 1, 1, 1))
272 << BatchNormalizationLayer(
273 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0b_3x3_BatchNorm_moving_mean.npy"),
274 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0b_3x3_BatchNorm_moving_variance.npy"),
275 get_random_accessor(1.f, 1.f),
276 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0b_3x3_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]277 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]278 << ConvolutionLayer(
279 3U, 3U, std::get<2>(c_filters),
280 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0c_3x3_weights.npy"),
281 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
282 PadStrideInfo(1, 1, 1, 1))
283 << BatchNormalizationLayer(
284 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0c_3x3_BatchNorm_moving_mean.npy"),
285 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0c_3x3_BatchNorm_moving_variance.npy"),
286 get_random_accessor(1.f, 1.f),
287 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0c_3x3_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]288 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU));
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]289
290 SubGraph i_d;
291 i_d << PoolingLayer(PoolingLayerInfo(PoolingType::AVG, 3, PadStrideInfo(1, 1, 1, 1, DimensionRoundingType::CEIL), true))
292 << ConvolutionLayer(
293 1U, 1U, d_filt,
294 get_weights_accessor(data_path, total_path + "Branch_3_Conv2d_0b_1x1_weights.npy"),
295 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
296 PadStrideInfo(1, 1, 0, 0))
297 << BatchNormalizationLayer(
298 get_weights_accessor(data_path, total_path + "Branch_3_Conv2d_0b_1x1_BatchNorm_moving_mean.npy"),
299 get_weights_accessor(data_path, total_path + "Branch_3_Conv2d_0b_1x1_BatchNorm_moving_variance.npy"),
300 get_random_accessor(1.f, 1.f),
301 get_weights_accessor(data_path, total_path + "Branch_3_Conv2d_0b_1x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]302 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU));
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]303
304 return BranchLayer(BranchMergeMethod::DEPTH_CONCATENATE, std::move(i_a), std::move(i_b), std::move(i_c), std::move(i_d));
305 }
306
307 BranchLayer get_inception_node_B(const std::string &data_path, std::string &&param_path,
308 unsigned int a_filt,
309 std::tuple<unsigned int, unsigned int, unsigned int> b_filters)
310 {
311 std::string total_path = "/cnn_data/inceptionv3_model/" + param_path + "_";
312 SubGraph i_a;
313 i_a << ConvolutionLayer(
314 3U, 3U, a_filt,
315 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_1a_1x1_weights.npy"),
316 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
317 PadStrideInfo(2, 2, 0, 0))
318 << BatchNormalizationLayer(
319 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_1a_1x1_BatchNorm_moving_mean.npy"),
320 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_1a_1x1_BatchNorm_moving_variance.npy"),
321 get_random_accessor(1.f, 1.f),
322 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_1a_1x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]323 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU));
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]324
325 SubGraph i_b;
326 i_b << ConvolutionLayer(
327 1U, 1U, std::get<0>(b_filters),
328 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0a_1x1_weights.npy"),
329 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
330 PadStrideInfo(1, 1, 0, 0))
331 << BatchNormalizationLayer(
332 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0a_1x1_BatchNorm_moving_mean.npy"),
333 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0a_1x1_BatchNorm_moving_variance.npy"),
334 get_random_accessor(1.f, 1.f),
335 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0a_1x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]336 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]337 << ConvolutionLayer(
338 3U, 3U, std::get<1>(b_filters),
339 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0b_3x3_weights.npy"),
340 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
341 PadStrideInfo(1, 1, 1, 1))
342 << BatchNormalizationLayer(
343 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0b_3x3_BatchNorm_moving_mean.npy"),
344 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0b_3x3_BatchNorm_moving_variance.npy"),
345 get_random_accessor(1.f, 1.f),
346 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0b_3x3_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]347 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]348 << ConvolutionLayer(
349 3U, 3U, std::get<2>(b_filters),
350 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_1a_1x1_weights.npy"),
351 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
352 PadStrideInfo(2, 2, 0, 0))
353 << BatchNormalizationLayer(
354 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_1a_1x1_BatchNorm_moving_mean.npy"),
355 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_1a_1x1_BatchNorm_moving_variance.npy"),
356 get_random_accessor(1.f, 1.f),
357 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_1a_1x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]358 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU));
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]359
360 SubGraph i_c;
361 i_c << PoolingLayer(PoolingLayerInfo(PoolingType::MAX, 3, PadStrideInfo(2, 2, 0, 0, DimensionRoundingType::CEIL)))
362 // TODO (geopin01) : Remove once we understand why a single node graph does not run in CL
363 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::LINEAR, 1.f, 0.f));
364
365 return BranchLayer(BranchMergeMethod::DEPTH_CONCATENATE, std::move(i_a), std::move(i_b), std::move(i_c));
366 }
367
368 BranchLayer get_inception_node_C(const std::string &data_path, std::string &&param_path,
369 unsigned int a_filt,
370 std::tuple<unsigned int, unsigned int, unsigned int> b_filters,
371 std::tuple<unsigned int, unsigned int, unsigned int, unsigned int, unsigned int> c_filters,
372 unsigned int d_filt)
373 {
374 std::string total_path = "/cnn_data/inceptionv3_model/" + param_path + "_";
375 SubGraph i_a;
376 i_a << ConvolutionLayer(
377 1U, 1U, a_filt,
378 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_0a_1x1_weights.npy"),
379 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
380 PadStrideInfo(1, 1, 0, 0))
381 << BatchNormalizationLayer(
382 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_0a_1x1_BatchNorm_moving_mean.npy"),
383 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_0a_1x1_BatchNorm_moving_variance.npy"),
384 get_random_accessor(1.f, 1.f),
385 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_0a_1x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]386 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU));
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]387
388 SubGraph i_b;
389 i_b << ConvolutionLayer(
390 1U, 1U, std::get<0>(b_filters),
391 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0a_1x1_weights.npy"),
392 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
393 PadStrideInfo(1, 1, 0, 0))
394 << BatchNormalizationLayer(
395 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0a_1x1_BatchNorm_moving_mean.npy"),
396 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0a_1x1_BatchNorm_moving_variance.npy"),
397 get_random_accessor(1.f, 1.f),
398 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0a_1x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]399 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]400 << ConvolutionLayer(
401 7U, 1U, std::get<1>(b_filters),
402 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0b_1x7_weights.npy"),
403 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
404 PadStrideInfo(1, 1, 3, 0))
405 << BatchNormalizationLayer(
406 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0b_1x7_BatchNorm_moving_mean.npy"),
407 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0b_1x7_BatchNorm_moving_variance.npy"),
408 get_random_accessor(1.f, 1.f),
409 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0b_1x7_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]410 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]411 << ConvolutionLayer(
412 1U, 7U, std::get<2>(b_filters),
413 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0c_7x1_weights.npy"),
414 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
415 PadStrideInfo(1, 1, 0, 3))
416 << BatchNormalizationLayer(
417 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0c_7x1_BatchNorm_moving_mean.npy"),
418 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0c_7x1_BatchNorm_moving_variance.npy"),
419 get_random_accessor(1.f, 1.f),
420 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0c_7x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]421 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU));
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]422
423 SubGraph i_c;
424 i_c << ConvolutionLayer(
425 1U, 1U, std::get<0>(c_filters),
426 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0a_1x1_weights.npy"),
427 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
428 PadStrideInfo(1, 1, 0, 0))
429 << BatchNormalizationLayer(
430 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0a_1x1_BatchNorm_moving_mean.npy"),
431 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0a_1x1_BatchNorm_moving_variance.npy"),
432 get_random_accessor(1.f, 1.f),
433 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0a_1x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]434 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]435 << ConvolutionLayer(
436 1U, 7U, std::get<1>(c_filters),
437 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0b_7x1_weights.npy"),
438 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
439 PadStrideInfo(1, 1, 0, 3))
440 << BatchNormalizationLayer(
441 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0b_7x1_BatchNorm_moving_mean.npy"),
442 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0b_7x1_BatchNorm_moving_variance.npy"),
443 get_random_accessor(1.f, 1.f),
444 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0b_7x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]445 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]446 << ConvolutionLayer(
447 7U, 1U, std::get<2>(c_filters),
448 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0c_1x7_weights.npy"),
449 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
450 PadStrideInfo(1, 1, 3, 0))
451 << BatchNormalizationLayer(
452 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0c_1x7_BatchNorm_moving_mean.npy"),
453 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0c_1x7_BatchNorm_moving_variance.npy"),
454 get_random_accessor(1.f, 1.f),
455 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0c_1x7_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]456 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]457 << ConvolutionLayer(
458 1U, 7U, std::get<3>(c_filters),
459 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0d_7x1_weights.npy"),
460 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
461 PadStrideInfo(1, 1, 0, 3))
462 << BatchNormalizationLayer(
463 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0d_7x1_BatchNorm_moving_mean.npy"),
464 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0d_7x1_BatchNorm_moving_variance.npy"),
465 get_random_accessor(1.f, 1.f),
466 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0d_7x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]467 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]468 << ConvolutionLayer(
469 7U, 1U, std::get<4>(c_filters),
470 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0e_1x7_weights.npy"),
471 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
472 PadStrideInfo(1, 1, 3, 0))
473 << BatchNormalizationLayer(
474 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0e_1x7_BatchNorm_moving_mean.npy"),
475 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0e_1x7_BatchNorm_moving_variance.npy"),
476 get_random_accessor(1.f, 1.f),
477 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0e_1x7_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]478 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU));
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]479
480 SubGraph i_d;
481 i_d << PoolingLayer(PoolingLayerInfo(PoolingType::AVG, 3, PadStrideInfo(1, 1, 1, 1, DimensionRoundingType::CEIL), true))
482 << ConvolutionLayer(
483 1U, 1U, d_filt,
484 get_weights_accessor(data_path, total_path + "Branch_3_Conv2d_0b_1x1_weights.npy"),
485 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
486 PadStrideInfo(1, 1, 0, 0))
487 << BatchNormalizationLayer(
488 get_weights_accessor(data_path, total_path + "Branch_3_Conv2d_0b_1x1_BatchNorm_moving_mean.npy"),
489 get_weights_accessor(data_path, total_path + "Branch_3_Conv2d_0b_1x1_BatchNorm_moving_variance.npy"),
490 get_random_accessor(1.f, 1.f),
491 get_weights_accessor(data_path, total_path + "Branch_3_Conv2d_0b_1x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]492 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU));
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]493
494 return BranchLayer(BranchMergeMethod::DEPTH_CONCATENATE, std::move(i_a), std::move(i_b), std::move(i_c), std::move(i_d));
495 }
496
497 BranchLayer get_inception_node_D(const std::string &data_path, std::string &&param_path,
498 std::tuple<unsigned int, unsigned int> a_filters,
499 std::tuple<unsigned int, unsigned int, unsigned int, unsigned int> b_filters)
500 {
501 std::string total_path = "/cnn_data/inceptionv3_model/" + param_path + "_";
502 SubGraph i_a;
503 i_a << ConvolutionLayer(
504 1U, 1U, std::get<0>(a_filters),
505 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_0a_1x1_weights.npy"),
506 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
507 PadStrideInfo(1, 1, 0, 0))
508 << BatchNormalizationLayer(
509 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_0a_1x1_BatchNorm_moving_mean.npy"),
510 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_0a_1x1_BatchNorm_moving_variance.npy"),
511 get_random_accessor(1.f, 1.f),
512 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_0a_1x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]513 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]514 << ConvolutionLayer(
515 3U, 3U, std::get<1>(a_filters),
516 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_1a_3x3_weights.npy"),
517 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
518 PadStrideInfo(2, 2, 0, 0))
519 << BatchNormalizationLayer(
520 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_1a_3x3_BatchNorm_moving_mean.npy"),
521 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_1a_3x3_BatchNorm_moving_variance.npy"),
522 get_random_accessor(1.f, 1.f),
523 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_1a_3x3_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]524 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU));
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]525
526 SubGraph i_b;
527 i_b << ConvolutionLayer(
528 1U, 1U, std::get<0>(b_filters),
529 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0a_1x1_weights.npy"),
530 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
531 PadStrideInfo(1, 1, 0, 0))
532 << BatchNormalizationLayer(
533 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0a_1x1_BatchNorm_moving_mean.npy"),
534 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0a_1x1_BatchNorm_moving_variance.npy"),
535 get_random_accessor(1.f, 1.f),
536 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0a_1x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]537 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]538 << ConvolutionLayer(
539 7U, 1U, std::get<1>(b_filters),
540 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0b_1x7_weights.npy"),
541 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
542 PadStrideInfo(1, 1, 3, 0))
543 << BatchNormalizationLayer(
544 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0b_1x7_BatchNorm_moving_mean.npy"),
545 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0b_1x7_BatchNorm_moving_variance.npy"),
546 get_random_accessor(1.f, 1.f),
547 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0b_1x7_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]548 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]549 << ConvolutionLayer(
550 1U, 7U, std::get<2>(b_filters),
551 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0c_7x1_weights.npy"),
552 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
553 PadStrideInfo(1, 1, 0, 3))
554 << BatchNormalizationLayer(
555 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0c_7x1_BatchNorm_moving_mean.npy"),
556 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0c_7x1_BatchNorm_moving_variance.npy"),
557 get_random_accessor(1.f, 1.f),
558 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0c_7x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]559 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]560 << ConvolutionLayer(
561 3U, 3U, std::get<3>(b_filters),
562 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_1a_3x3_weights.npy"),
563 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
564 PadStrideInfo(2, 2, 0, 0))
565 << BatchNormalizationLayer(
566 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_1a_3x3_BatchNorm_moving_mean.npy"),
567 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_1a_3x3_BatchNorm_moving_variance.npy"),
568 get_random_accessor(1.f, 1.f),
569 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_1a_3x3_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]570 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU));
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]571
572 SubGraph i_c;
573 i_c << PoolingLayer(PoolingLayerInfo(PoolingType::MAX, 3, PadStrideInfo(2, 2, 0, 0, DimensionRoundingType::CEIL)))
574 // TODO (geopin01) : Remove once we understand why a single node graph does not run in CL
575 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::LINEAR, 1.f, 0.f));
576
577 return BranchLayer(BranchMergeMethod::DEPTH_CONCATENATE, std::move(i_a), std::move(i_b), std::move(i_c));
578 }
579
580 BranchLayer get_inception_node_E(const std::string &data_path, std::string &&param_path,
581 unsigned int a_filt,
582 std::tuple<unsigned int, unsigned int, unsigned int> b_filters,
583 std::tuple<unsigned int, unsigned int, unsigned int, unsigned int> c_filters,
584 unsigned int d_filt,
585 bool is_name_different = false)
586 {
587 // This is due to a naming issue in the tf model
588 std::string conv_id = "_0b_";
589 if(is_name_different)
590 {
591 conv_id = "_0c_";
592 }
593
594 std::string total_path = "/cnn_data/inceptionv3_model/" + param_path + "_";
595 SubGraph i_a;
596 i_a << ConvolutionLayer(
597 1U, 1U, a_filt,
598 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_0a_1x1_weights.npy"),
599 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
600 PadStrideInfo(1, 1, 0, 0))
601 << BatchNormalizationLayer(
602 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_0a_1x1_BatchNorm_moving_mean.npy"),
603 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_0a_1x1_BatchNorm_moving_variance.npy"),
604 get_random_accessor(1.f, 1.f),
605 get_weights_accessor(data_path, total_path + "Branch_0_Conv2d_0a_1x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]606 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU));
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]607
608 SubGraph i_b1;
609 i_b1 << ConvolutionLayer(
610 3U, 1U, std::get<1>(b_filters),
611 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0b_1x3_weights.npy"),
612 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
613 PadStrideInfo(1, 1, 1, 0))
614 << BatchNormalizationLayer(
615 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0b_1x3_BatchNorm_moving_mean.npy"),
616 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0b_1x3_BatchNorm_moving_variance.npy"),
617 get_random_accessor(1.f, 1.f),
618 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0b_1x3_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]619 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU));
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]620
621 SubGraph i_b2;
622 i_b2 << ConvolutionLayer(
623 1U, 3U, std::get<2>(b_filters),
624 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d" + conv_id + "3x1_weights.npy"),
625 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
626 PadStrideInfo(1, 1, 0, 1))
627 << BatchNormalizationLayer(
628 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d" + conv_id + "3x1_BatchNorm_moving_mean.npy"),
629 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d" + conv_id + "3x1_BatchNorm_moving_variance.npy"),
630 get_random_accessor(1.f, 1.f),
631 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d" + conv_id + "3x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]632 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU));
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]633
634 SubGraph i_b;
635 i_b << ConvolutionLayer(
636 1U, 1U, std::get<0>(b_filters),
637 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0a_1x1_weights.npy"),
638 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
639 PadStrideInfo(1, 1, 0, 0))
640 << BatchNormalizationLayer(
641 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0a_1x1_BatchNorm_moving_mean.npy"),
642 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0a_1x1_BatchNorm_moving_variance.npy"),
643 get_random_accessor(1.f, 1.f),
644 get_weights_accessor(data_path, total_path + "Branch_1_Conv2d_0a_1x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]645 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]646 << BranchLayer(BranchMergeMethod::DEPTH_CONCATENATE, std::move(i_b1), std::move(i_b2));
647
648 SubGraph i_c1;
649 i_c1 << ConvolutionLayer(
650 3U, 1U, std::get<2>(c_filters),
651 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0c_1x3_weights.npy"),
652 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
653 PadStrideInfo(1, 1, 1, 0))
654 << BatchNormalizationLayer(
655 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0c_1x3_BatchNorm_moving_mean.npy"),
656 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0c_1x3_BatchNorm_moving_variance.npy"),
657 get_random_accessor(1.f, 1.f),
658 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0c_1x3_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]659 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU));
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]660
661 SubGraph i_c2;
662 i_c2 << ConvolutionLayer(
663 1U, 3U, std::get<3>(c_filters),
664 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0d_3x1_weights.npy"),
665 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
666 PadStrideInfo(1, 1, 0, 1))
667 << BatchNormalizationLayer(
668 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0d_3x1_BatchNorm_moving_mean.npy"),
669 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0d_3x1_BatchNorm_moving_variance.npy"),
670 get_random_accessor(1.f, 1.f),
671 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0d_3x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]672 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU));
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]673
674 SubGraph i_c;
675 i_c << ConvolutionLayer(
676 1U, 1U, std::get<0>(c_filters),
677 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0a_1x1_weights.npy"),
678 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
679 PadStrideInfo(1, 1, 0, 0))
680 << BatchNormalizationLayer(
681 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0a_1x1_BatchNorm_moving_mean.npy"),
682 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0a_1x1_BatchNorm_moving_variance.npy"),
683 get_random_accessor(1.f, 1.f),
684 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0a_1x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]685 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]686 << ConvolutionLayer(
687 3U, 3U, std::get<1>(c_filters),
688 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0b_3x3_weights.npy"),
689 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
690 PadStrideInfo(1, 1, 1, 1))
691 << BatchNormalizationLayer(
692 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0b_3x3_BatchNorm_moving_mean.npy"),
693 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0b_3x3_BatchNorm_moving_variance.npy"),
694 get_random_accessor(1.f, 1.f),
695 get_weights_accessor(data_path, total_path + "Branch_2_Conv2d_0b_3x3_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]696 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]697 << BranchLayer(BranchMergeMethod::DEPTH_CONCATENATE, std::move(i_c1), std::move(i_c2));
698
699 SubGraph i_d;
700 i_d << PoolingLayer(PoolingLayerInfo(PoolingType::AVG, 3, PadStrideInfo(1, 1, 1, 1, DimensionRoundingType::CEIL), true))
701 << ConvolutionLayer(
702 1U, 1U, d_filt,
703 get_weights_accessor(data_path, total_path + "Branch_3_Conv2d_0b_1x1_weights.npy"),
704 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
705 PadStrideInfo(1, 1, 0, 0))
706 << BatchNormalizationLayer(
707 get_weights_accessor(data_path, total_path + "Branch_3_Conv2d_0b_1x1_BatchNorm_moving_mean.npy"),
708 get_weights_accessor(data_path, total_path + "Branch_3_Conv2d_0b_1x1_BatchNorm_moving_variance.npy"),
709 get_random_accessor(1.f, 1.f),
710 get_weights_accessor(data_path, total_path + "Branch_3_Conv2d_0b_1x1_BatchNorm_beta.npy"),
Giorgio Arena11674872018-02-07 15:38:12 +0000[diff] [blame]711 0.001f, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU));
Georgios Pinitas652bde52018-01-10 15:33:28 +0000[diff] [blame]712
713 return BranchLayer(BranchMergeMethod::DEPTH_CONCATENATE, std::move(i_a), std::move(i_b), std::move(i_c), std::move(i_d));
714 }
715};
716
717/** Main program for Inception V3
718 *
719 * @param[in] argc Number of arguments
720 * @param[in] argv Arguments ( [optional] Target (0 = NEON, 1 = OpenCL), [optional] Path to the weights folder, [optional] image, [optional] labels )
721 */
722int main(int argc, char **argv)
723{
724 return arm_compute::utils::run_example<InceptionV3Example>(argc, argv);
725}