COMPMID-926 Add depth multiplier support to NEON/CL/GLES depthwise convolution

Change-Id: I03f32c62350e5ea43e77bb15fc5a832d83719e3b
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/126657
Tested-by: Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele DiGiorgio <michele.digiorgio@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>

1 files changed, 10 insertions, 9 deletions

diff --git a/arm_compute/core/CL/kernels/CLDepthwiseIm2ColKernel.h b/arm_compute/core/CL/kernels/CLDepthwiseIm2ColKernel.h
index 86b8b90e94..b8343da50a 100644
--- a/arm_compute/core/CL/kernels/CLDepthwiseIm2ColKernel.h
+++ b/arm_compute/core/CL/kernels/CLDepthwiseIm2ColKernel.h
@@ -33,7 +33,7 @@ class ICLTensor;
 
 /** Interface for the depthwise im2col reshape kernel.
  *  This kernel reshape the input low 3 dimensions to a new 3D shape  where the output's first dimension is
- *  the linear patch size (FILTER_WIDTH * FILTER_HEIGHT) and second dimension is number of patches in per image and third dimension unchanged .
+ *  the linear patch size (FILTER_WIDTH * FILTER_HEIGHT) and second dimension is number of patches per image and third dimension unchanged .
  **/
 class CLDepthwiseIm2ColKernel : public ICLKernel
 {
@@ -50,15 +50,16 @@ public:
     CLDepthwiseIm2ColKernel &operator=(CLDepthwiseIm2ColKernel &&) = default;
     /** Set the input and output of the kernel.
      *
-     * @param[in]  input       The input tensor to convert. 3 lower dimensions represent a single input [width, height, IFM],
-     *                         while every optional dimension from 4 and above represent a batch of inputs. Data types supported: QASYMM8/F32
-     * @param[out] output      The output tensor. First 3 lower dimensions represent a transform of each 3D input,
-     *                         while every dimension above 3 represents a batch. Data types supported: Same as @p input
-     * @param[in]  kernel_dims The kernel dimensions (width and height).
-     * @param[in]  conv_info   Contains padding and stride information described in @ref PadStrideInfo.
-     * @param[in]  has_bias    Boolean that specifies if the depthwise convolution has bias.
+     * @param[in]  input            The input tensor to convert. 3 lower dimensions represent a single input [width, height, IFM],
+     *                              while every optional dimension from 4 and above represent a batch of inputs. Data types supported: QASYMM8/F32
+     * @param[out] output           The output tensor. First 3 lower dimensions represent a transform of each 3D input,
+     *                              while every dimension above 3 represents a batch. Data types supported: Same as @p input
+     * @param[in]  kernel_dims      The kernel dimensions (width and height).
+     * @param[in]  conv_info        Contains padding and stride information described in @ref PadStrideInfo.
+     * @param[in]  has_bias         Boolean that specifies if the depthwise convolution has bias.
+     * @param[in]  depth_multiplier (Optional) Multiplier to apply to the input's depth in order to retrieve the output's depth. Defaults to 1.
      */
-    void configure(const ICLTensor *input, ICLTensor *output, const Size2D &kernel_dims, const PadStrideInfo &conv_info, bool has_bias = false);
+    void configure(const ICLTensor *input, ICLTensor *output, const Size2D &kernel_dims, const PadStrideInfo &conv_info, bool has_bias = false, unsigned int depth_multiplier = 1);
 
     // Inherited methods overridden:
     void run(const Window &window, cl::CommandQueue &queue) override;