From 894066de8cc26d1a3aca62dcaa6b30a2a1116028 Mon Sep 17 00:00:00 2001 From: George Wort Date: Fri, 15 Feb 2019 15:12:52 +0000 Subject: COMPMID-1844: Implement CLCrop Change-Id: I8822c37adc45960705dc3f32a53214795ba3cf39 Signed-off-by: George Wort Reviewed-on: https://review.mlplatform.org/c/789 Reviewed-by: Manuel Bottini Tested-by: Arm Jenkins Reviewed-by: Pablo Marquez --- arm_compute/core/CL/CLKernels.h | 1 + arm_compute/core/CL/kernels/CLCopyKernel.h | 22 ++++--- arm_compute/core/CL/kernels/CLCropKernel.h | 88 ++++++++++++++++++++++++++++ arm_compute/core/CL/kernels/CLMemsetKernel.h | 9 ++- 4 files changed, 108 insertions(+), 12 deletions(-) create mode 100644 arm_compute/core/CL/kernels/CLCropKernel.h (limited to 'arm_compute/core/CL') diff --git a/arm_compute/core/CL/CLKernels.h b/arm_compute/core/CL/CLKernels.h index e0d124bff9..396158ff2d 100644 --- a/arm_compute/core/CL/CLKernels.h +++ b/arm_compute/core/CL/CLKernels.h @@ -46,6 +46,7 @@ #include "arm_compute/core/CL/kernels/CLConvertFullyConnectedWeightsKernel.h" #include "arm_compute/core/CL/kernels/CLConvolutionKernel.h" #include "arm_compute/core/CL/kernels/CLCopyKernel.h" +#include "arm_compute/core/CL/kernels/CLCropKernel.h" #include "arm_compute/core/CL/kernels/CLDeconvolutionLayerUpsampleKernel.h" #include "arm_compute/core/CL/kernels/CLDepthConcatenateLayerKernel.h" #include "arm_compute/core/CL/kernels/CLDepthConvertLayerKernel.h" diff --git a/arm_compute/core/CL/kernels/CLCopyKernel.h b/arm_compute/core/CL/kernels/CLCopyKernel.h index 9fc3759f4e..464fa13791 100644 --- a/arm_compute/core/CL/kernels/CLCopyKernel.h +++ b/arm_compute/core/CL/kernels/CLCopyKernel.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2018 ARM Limited. + * Copyright (c) 2018-2019 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -47,20 +47,22 @@ public: CLCopyKernel &operator=(CLCopyKernel &&) = default; /** Initialize the kernel's input, output. * - * @param[in] input Source tensor. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32. - * @param[out] output Destination tensor. Data types supported: same as @p input. - * @param[in] padding (Optional) Padding to be applied to the input tensor + * @param[in] input Source tensor. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32. + * @param[out] output Destination tensor. Data types supported: same as @p input. + * @param[in] padding (Optional) Padding to be applied to the input tensor + * @param[in] output_window (Optional) Window to be used in case only copying into part of a tensor. Default is nullptr. */ - void configure(const ICLTensor *input, ICLTensor *output, const PaddingList &padding = PaddingList()); + void configure(const ICLTensor *input, ICLTensor *output, const PaddingList &padding = PaddingList(), Window *output_window = nullptr); /** Static function to check if given info will lead to a valid configuration of @ref CLCopyKernel * - * @param[in] input Source tensor info. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32. - * @param[in] output Destination tensor info. Data types supported: same as @p input. - * @param[in] padding (Optional) Padding to be applied to the input tensor + * @param[in] input Source tensor info. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32. + * @param[in] output Destination tensor info. Data types supported: same as @p input. + * @param[in] padding (Optional) Padding to be applied to the input tensor + * @param[in] output_window (Optional) Window to be used in case only copying into part of a tensor. Default is nullptr. * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, const PaddingList &padding = PaddingList()); + static Status validate(const ITensorInfo *input, const ITensorInfo *output, const PaddingList &padding = PaddingList(), Window *output_window = nullptr); // Inherited methods overridden: void run(const Window &window, cl::CommandQueue &queue) override; @@ -68,6 +70,8 @@ public: private: const ICLTensor *_input; ICLTensor *_output; + Window _output_window; + bool _has_output_window; }; } // namespace arm_compute #endif /*__ARM_COMPUTE_CLCOPYKERNEL_H__ */ diff --git a/arm_compute/core/CL/kernels/CLCropKernel.h b/arm_compute/core/CL/kernels/CLCropKernel.h new file mode 100644 index 0000000000..ff4b13f865 --- /dev/null +++ b/arm_compute/core/CL/kernels/CLCropKernel.h @@ -0,0 +1,88 @@ +/* + * Copyright (c) 2019 ARM Limited. + * + * SPDX-License-Identifier: MIT + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to + * deal in the Software without restriction, including without limitation the + * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or + * sell copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in all + * copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ +#ifndef __ARM_COMPUTE_CLCROPKERNEL_H__ +#define __ARM_COMPUTE_CLCROPKERNEL_H__ + +#include "arm_compute/core/CL/ICLKernel.h" +#include "arm_compute/core/Types.h" + +namespace arm_compute +{ +class ICLTensor; + +/** OpenCL kernel to perform a copy between two tensors */ +class CLCropKernel : public ICLKernel +{ +public: + /** Default constructor */ + CLCropKernel(); + /** Prevent instances of this class from being copied (As this class contains pointers). */ + CLCropKernel(const CLCropKernel &) = delete; + /** Prevent instances of this class from being copied (As this class contains pointers). */ + CLCropKernel &operator=(const CLCropKernel &) = delete; + /** Allow instances of this class to be moved */ + CLCropKernel(CLCropKernel &&) = default; + /** Allow instances of this class to be moved */ + CLCropKernel &operator=(CLCropKernel &&) = default; + /** Configure kernel + * + * @note Supported tensor rank: up to 4 + * + * @param[in] input Source tensor. Data type supported: U16/S16/U32/S32/F16/F32. Data layouts supported: NHWC. + * @param[out] output Destination tensor. Data type supported: F32 + * @param[in] start Coordinates of where to start cropping the image. + * @param[in] end Coordinates of where to end cropping the image. + * @param[in] batch_index Fourth dimension index of the 3D image to crop in @p input. + * @param[in] extrapolation_value Value to be used for values outside of the image. Default is 0. + * @param[in] output_window Output window to be used in case cropped image is being copied into a tensor. Default is nullptr. + */ + void configure(const ICLTensor *input, ICLTensor *output, Coordinates2D start, Coordinates2D end, uint32_t batch_index, float extrapolation_value = 0, Window *output_window = nullptr); + + /** Static function to check if given info will lead to a valid configuration of @ref CLStridedSliceKernel + * + * @note Supported tensor rank: up to 4 + * + * @param[in] input Source tensor info. Data type supported: U16/S16/U32/S32/F16/F32. Data layouts supported: NHWC. + * @param[in] output Destination tensor info. Data type supported: F32 + * @param[in] start Coordinates of where to start cropping the image. + * @param[in] end Coordinates of where to end cropping the image. + * @param[in] batch_index Fourth dimension index of the 3D image to crop in @p input. + * @param[in] extrapolation_value Value to be used for values outside of the image. Default is 0. + * @param[in] output_window Output window to be used in case cropped image is being copied into a tensor. Default is nullptr. + */ + static Status validate(const ITensorInfo *input, const ITensorInfo *output, Coordinates2D start, Coordinates2D end, uint32_t batch_index, float extrapolation_value = 0, + Window *output_window = nullptr); + + // Inherited methods overridden: + void run(const Window &window, cl::CommandQueue &queue) override; + +private: + const ICLTensor *_input; + ICLTensor *_output; + Coordinates2D _start; + uint32_t _batch_index; + float _extrapolation_value; +}; +} // namespace arm_compute +#endif /*__ARM_COMPUTE_CLCROPKERNEL_H__ */ diff --git a/arm_compute/core/CL/kernels/CLMemsetKernel.h b/arm_compute/core/CL/kernels/CLMemsetKernel.h index e35a9f1cf2..c79b6f4d60 100644 --- a/arm_compute/core/CL/kernels/CLMemsetKernel.h +++ b/arm_compute/core/CL/kernels/CLMemsetKernel.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2018 ARM Limited. + * Copyright (c) 2018-2019 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -53,22 +53,25 @@ public: * * @param[in,out] tensor Input tensor to fill. Supported data types: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32 * @param[in] constant_value The value used to fill the planes of the tensor + * @param[in] window Window to be used in case setting only part of a tensor. Default is nullptr. */ - void configure(ICLTensor *tensor, const PixelValue &constant_value); + void configure(ICLTensor *tensor, const PixelValue &constant_value, Window *window = nullptr); /** Static function to check if given info will lead to a valid configuration of @ref CLMemsetKernel * * @param[in] tensor Source tensor info. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32 * @param[in] constant_value The value used to fill the planes of the tensor + * @param[in] window Window to be used in case setting only part of a tensor. Default is nullptr. * * @return a status */ - static Status validate(const ITensorInfo *tensor, const PixelValue &constant_value); + static Status validate(const ITensorInfo *tensor, const PixelValue &constant_value, Window *window = nullptr); // Inherited methods overridden: void run(const Window &window, cl::CommandQueue &queue) override; private: ICLTensor *_tensor; + Window _full_window; }; } // namespace arm_compute #endif /*__ARM_COMPUTE_CLMEMSETRKERNEL_H__ */ -- cgit v1.2.1