commit 0c2eeac6347533a1d3d456aebea492f5123388f3
author Keith Davis <keith.davis@arm.com> Tue Feb 11 16:51:50 2020 +0000
committer James Conroy <james.conroy@arm.com> Mon Feb 17 21:53:29 2020 +0000
tree f218fc236137791c491b680dfd24fb9706c171a6
parent 4c3c1f486ab775eacb1f6455f8468f9be2c3e4f7

IVGCVSW-4436 Add ExecuteNetwork test for mobilenet_v2_int8

 * Add QAsymmS8 to QueueDescriptor supportedTypes
 * Add QSymmS8/QAsymmS8 to RefLayerSupport supportedTypes
 * Some additional comments and refactoring

Change-Id: I8567314452e6e8f6f69cb6e458ee147d3fc92fab
Signed-off-by: Keith Davis <keith.davis@arm.com>

src/armnnTfLiteParser/TfLiteParser.cpp

diff --git a/src/armnnTfLiteParser/TfLiteParser.cpp b/src/armnnTfLiteParser/TfLiteParser.cpp
index 560cdf1..593f3eb 100644
--- a/src/armnnTfLiteParser/TfLiteParser.cpp
+++ b/src/armnnTfLiteParser/TfLiteParser.cpp
@@ -301,7 +301,8 @@
     }
 }
 
-armnn::TensorInfo ToTensorInfo(TfLiteParser::TensorRawPtr tensorPtr, const std::vector<unsigned int>& shapes)
+armnn::TensorInfo ToTensorInfo(TfLiteParser::TensorRawPtr tensorPtr, const std::vector<unsigned int>& shapes,
+                               const armnn::PermutationVector& dimensionMappings = {0, 1, 2, 3})
 {
     armnn::DataType type;
     CHECK_TENSOR_PTR(tensorPtr);
@@ -317,10 +318,12 @@
         case tflite::TensorType_INT8:
             if (tensorPtr->quantization->zero_point.size() == 1 && tensorPtr->quantization->zero_point[0] != 0)
             {
+                // Per-tensor
                 type = armnn::DataType::QAsymmS8;
             }
             else
             {
+                // Per-channel
                 type = armnn::DataType::QSymmS8;
             }
             break;
@@ -388,12 +391,13 @@
                       tensorPtr->quantization->scale.end(),
                       std::back_inserter(quantizationScales));
 
-            // QSymm Per-axis
+            // QSymmS8 Per-axis
             armnn::TensorInfo result(boost::numeric_cast<unsigned int>(safeShape.size()),
                               safeShape.data(),
                               type,
                               quantizationScales,
-                              boost::numeric_cast<unsigned int>(tensorPtr->quantization->quantized_dimension));
+                              dimensionMappings[boost::numeric_cast<unsigned int>(
+                              tensorPtr->quantization->quantized_dimension)]);
 
             return result;
         }
@@ -409,10 +413,11 @@
     }
 }
 
-armnn::TensorInfo ToTensorInfo(TfLiteParser::TensorRawPtr tensorPtr)
+armnn::TensorInfo ToTensorInfo(TfLiteParser::TensorRawPtr tensorPtr, 
+                               const armnn::PermutationVector& dimensionMappings = {0, 1, 2, 3})
 {
     auto const & dimensions = AsUnsignedVector(tensorPtr->shape);
-    return ToTensorInfo(tensorPtr, dimensions);
+    return ToTensorInfo(tensorPtr, dimensions, dimensionMappings);
 }
 
 template<typename T>
@@ -905,8 +910,11 @@
     desc.m_DilationX = CHECKED_NON_NEGATIVE(options->dilation_w_factor);
     desc.m_DilationY = CHECKED_NON_NEGATIVE(options->dilation_h_factor);
 
+    // Mappings from TensorflowLite filter tensors to the ArmNN filter tensors (ArmNN weights have to be [M, I, H, W])
+    PermutationVector permutationVector{ 2, 3, 1, 0 }; // [H, W, I, M] -> [M, I, H, W]
+  
     armnn::TensorInfo inputTensorInfo  = ToTensorInfo(inputs[0]);
-    armnn::TensorInfo filterTensorInfo = ToTensorInfo(inputs[1]);
+    armnn::TensorInfo filterTensorInfo = ToTensorInfo(inputs[1], permutationVector);
 
     // Assuming input is NHWC
     unsigned int inputHeight = inputTensorInfo.GetShape()[1];
@@ -922,9 +930,6 @@
                                 inputTensorInfo.GetShape()[3],
                                 filterTensorInfo.GetShape()[3] / inputTensorInfo.GetShape()[3] });
 
-    // Mappings from TensorflowLite filter tensors to the ArmNN filter tensors (ArmNN weights have to be [M, I, H, W])
-    PermutationVector permutationVector{ 2, 3, 1, 0 }; // [H, W, I, M] -> [M, I, H, W]
-
     CalcPadding(inputHeight, filterHeight, desc.m_StrideY,
                 desc.m_DilationY, desc.m_PadTop, desc.m_PadBottom, options->padding);
     CalcPadding(inputWidth, filterWidth, desc.m_StrideX,