Use asPadstrideInfo utility to calc pad values in TransposeConv op (#3705)

This patch removes the internal calc of pad values during execution and uses asPadstrideInfo
utility to calc the same.

Signed-off-by: shubham <shub98.gupta@samsung.com>

diff --git a/runtimes/pure_arm_compute/src/compilation.cc b/runtimes/pure_arm_compute/src/compilation.cc
index aa88e82..63088bb 100644 (file)
--- a/runtimes/pure_arm_compute/src/compilation.cc
+++ b/runtimes/pure_arm_compute/src/compilation.cc
@@ -3953,8 +3953,8 @@ void Planner::visit(const ::internal::tflite::op::TransposeConv::Node &node)
     int ofm_index;
     int ifm_index;
     int ker_index;
-    int stride_width;
-    int stride_height;
+    Padding padding;
+    Stride stride;
   };
 
   Param param;
@@ -3963,8 +3963,12 @@ void Planner::visit(const ::internal::tflite::op::TransposeConv::Node &node)
   param.ifm_index = ifm_index.asInt();
   param.ker_index = ker_index.asInt();
 
-  param.stride_width = hstride;
-  param.stride_height = vstride;
+  param.stride.horizontal = hstride;
+  param.stride.vertical = vstride;
+
+  param.padding = (padding_type == ANEURALNETWORKS_PADDING_SAME)
+                      ? same_padding(ifm_shape, param.stride, ker_shape.W, ker_shape.H)
+                      : valid_padding();
 
   auto stage = [param](const IAllocationContext &ctx, IExecutionBuilder &builder) {
     auto ofm_alloc = ctx.at(::internal::tflite::operand::Index{param.ofm_index});
@@ -3975,8 +3979,10 @@ void Planner::visit(const ::internal::tflite::op::TransposeConv::Node &node)
 
     // Only rank 4 is supported
     const int rank = 4;
-    fn->configure(ifm_alloc, ker_alloc, ofm_alloc, param.stride_width, param.stride_height,
-                  getARMComputeAxises(rank));
+
+    auto tconv_info = asPadStringInfo(param.padding, param.stride);
+
+    fn->configure(ifm_alloc, ker_alloc, ofm_alloc, tconv_info, getARMComputeAxises(rank));
 
     builder.append("TransposeConv", std::move(fn));
   };

diff --git a/runtimes/pure_arm_compute/src/internal/layers/SimpleTransposeConv.cc b/runtimes/pure_arm_compute/src/internal/layers/SimpleTransposeConv.cc
index 12fb5eb..85587c4 100644 (file)
--- a/runtimes/pure_arm_compute/src/internal/layers/SimpleTransposeConv.cc
+++ b/runtimes/pure_arm_compute/src/internal/layers/SimpleTransposeConv.cc
@@ -18,8 +18,9 @@
 #include <arm_compute/runtime/CL/CLScheduler.h>
 
 void SimpleTransposeConv::configure(::arm_compute::ITensor *input, ::arm_compute::ITensor *weights,
-                                    ::arm_compute::ITensor *output, int32_t stride_width,
-                                    int32_t stride_height, ::arm_compute::Coordinates axises)
+                                    ::arm_compute::ITensor *output,
+                                    ::arm_compute::PadStrideInfo &tconv_info,
+                                    ::arm_compute::Coordinates axises)
 {
   auto rank = axises.num_dimensions();
 
@@ -28,20 +29,20 @@ void SimpleTransposeConv::configure(::arm_compute::ITensor *input, ::arm_compute
   _input = input;
   _weights = weights;
   _output = output;
-  _stride_width = stride_width;
-  _stride_height = stride_height;
+  _stride_width = tconv_info.stride().first;
+  _stride_height = tconv_info.stride().second;
+  _pad_width = tconv_info.pad_left();
+  _pad_height = tconv_info.pad_top();
   _axises = axises;
 }
 
 template <typename T>
-inline void ApplyTransposeConv(const ::arm_compute::TensorShape &input_shape,
-                               const ::arm_compute::ITensor *input_data,
-                               const ::arm_compute::TensorShape &filter_shape,
-                               const ::arm_compute::ITensor *filter_data,
-                               const ::arm_compute::TensorShape &output_shape,
-                               const ::arm_compute::ITensor *output_data,
-                               const int32_t stride_width, const int32_t stride_height,
-                               const ::arm_compute::Coordinates axises)
+inline void ApplyTransposeConv(
+    const ::arm_compute::TensorShape &input_shape, const ::arm_compute::ITensor *input_data,
+    const ::arm_compute::TensorShape &filter_shape, const ::arm_compute::ITensor *filter_data,
+    const ::arm_compute::TensorShape &output_shape, const ::arm_compute::ITensor *output_data,
+    const int32_t stride_width, const int32_t stride_height, const int32_t pad_width,
+    const int32_t pad_height, const ::arm_compute::Coordinates axises)
 {
   const int batches = input_shape[axises[0]];
   const int input_height = input_shape[axises[1]];
@@ -55,9 +56,6 @@ inline void ApplyTransposeConv(const ::arm_compute::TensorShape &input_shape,
   const int output_width = output_shape[axises[2]];
   const int output_depth = output_shape[axises[3]];
 
-  const int pad_width = (stride_width * (input_width - 1) + filter_width - output_width) / 2;
-  const int pad_height = (stride_height * (input_height - 1) + filter_height - output_height) / 2;
-
   // Although transpose convolution simplifies to convolution with transposed
   // weights for strides of 1, non-unitary striding complicates matters. To
   // keep this reference implementation as clear as possible, we use a
@@ -126,14 +124,16 @@ void SimpleTransposeConv::run()
   switch (_input->info()->data_type())
   {
     case ::arm_compute::DataType::S32:
-      ApplyTransposeConv<int32_t>(
-          _input->info()->tensor_shape(), _input, _weights->info()->tensor_shape(), _weights,
-          _output->info()->tensor_shape(), _output, _stride_width, _stride_height, _axises);
+      ApplyTransposeConv<int32_t>(_input->info()->tensor_shape(), _input,
+                                  _weights->info()->tensor_shape(), _weights,
+                                  _output->info()->tensor_shape(), _output, _stride_width,
+                                  _stride_height, _pad_width, _pad_height, _axises);
       break;
     case ::arm_compute::DataType::F32:
-      ApplyTransposeConv<float>(
-          _input->info()->tensor_shape(), _input, _weights->info()->tensor_shape(), _weights,
-          _output->info()->tensor_shape(), _output, _stride_width, _stride_height, _axises);
+      ApplyTransposeConv<float>(_input->info()->tensor_shape(), _input,
+                                _weights->info()->tensor_shape(), _weights,
+                                _output->info()->tensor_shape(), _output, _stride_width,
+                                _stride_height, _pad_width, _pad_height, _axises);
       break;
     default:
       ARM_COMPUTE_ERROR("DataType not supported");

diff --git a/runtimes/pure_arm_compute/src/internal/layers/SimpleTransposeConv.h b/runtimes/pure_arm_compute/src/internal/layers/SimpleTransposeConv.h
index bd18bcb..c551982 100644 (file)
--- a/runtimes/pure_arm_compute/src/internal/layers/SimpleTransposeConv.h
+++ b/runtimes/pure_arm_compute/src/internal/layers/SimpleTransposeConv.h
@@ -24,7 +24,8 @@ class SimpleTransposeConv : public ::arm_compute::IFunction
 {
 public:
   SimpleTransposeConv()
-      : _input(nullptr), _weights(nullptr), _output(nullptr), _stride_width(0), _stride_height(0)
+      : _input(nullptr), _weights(nullptr), _output(nullptr), _stride_width(0), _stride_height(0),
+        _pad_width(0), _pad_height(0)
   {
     // DO NOTHING
   }
@@ -34,12 +35,12 @@ public:
    * @param[in]  input              First tensor input.
    * @param[in]  weights            Weights
    * @param[out] output             Output tensor.
-   * @param[in]  stride_width       Stride width value
-   * @param[in]  stride_height      Stride height value
+   * @param[in]  tc_info            Contains padding and policies to be used in the deconvolution,
+   *                                this is decribed in @ref PadStrideInfo.
    * @param[in]  axises             Axises of rank 4
    */
   void configure(::arm_compute::ITensor *input, ::arm_compute::ITensor *weights,
-                 ::arm_compute::ITensor *output, int32_t stride_width, int32_t stride_height,
+                 ::arm_compute::ITensor *output, ::arm_compute::PadStrideInfo &tconv_info,
                  ::arm_compute::Coordinates axises = getARMComputeAxises(4));
 
   void run() override;
@@ -50,6 +51,8 @@ private:
   ::arm_compute::ITensor *_output;
   int32_t _stride_width;
   int32_t _stride_height;
+  int32_t _pad_width;
+  int32_t _pad_height;
   ::arm_compute::Coordinates _axises;
 };