1 /* 2 * Copyright (c) 2019-2021 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 #ifndef ARM_COMPUTE_CLDIRECTDECONVOLUTIONLAYER_H 25 #define ARM_COMPUTE_CLDIRECTDECONVOLUTIONLAYER_H 26 27 #include "arm_compute/runtime/CL/functions/CLConvolutionLayer.h" 28 #include "arm_compute/runtime/CL/functions/CLDeconvolutionLayerUpsample.h" 29 #include "arm_compute/runtime/CL/functions/CLReverse.h" 30 #include "arm_compute/runtime/CL/functions/CLTranspose.h" 31 32 #include "arm_compute/runtime/CL/CLTensor.h" 33 #include "arm_compute/runtime/IFunction.h" 34 #include "arm_compute/runtime/IMemoryManager.h" 35 #include "arm_compute/runtime/MemoryGroup.h" 36 37 #include <memory> 38 39 namespace arm_compute 40 { 41 class ICLTensor; 42 /** Function to run the deconvolution layer. 43 * 44 * Deconvolution Layer is the backward pass of Convolution Layer. First we transform the input depending on the stride and pad info and then perform a 1x1 45 * convolution pass. Input stride defines how many zeroes we should put between each element of the input and pad is the amount of padding. 46 * 47 * The relation between input to output is as follows: 48 * \f[ 49 * width\_output = (width\_input - 1) \cdot stride\_x - 2 \cdot padding\_x + kernel\_x 50 * \f] 51 * \f[ 52 * height\_output = (height\_input - 1) \cdot stride\_y - 2 \cdot padding\_y + kernel\_y 53 * \f] 54 * 55 * where: 56 * width_input is the size of the first input dimension. 57 * height_input is the size of the second input dimension. 58 * width_output is the size of the first output dimension. 59 * height_output is the size of the second output dimension. 60 * kernel_x and kernel_y are the convolution sizes in x and y. 61 * stride_x and stride_y is the input stride of the first and second dimension. 62 * 63 * The weights used by Deconvolution are supposed to be the same as the ones used for Convolution. Therefore, it will be necessary to use the weights in the 64 * reverse order to perform an actual convolution. This is achieved by using @ref CLReverse. 65 * 66 * This function calls the following OpenCL kernels/functions: 67 * 68 * -# @ref CLDeconvolutionLayerUpsample 69 * -# @ref CLConvolutionLayer 70 * 71 * And the following CPP kernels: 72 * -# @ref CLReverse 73 * 74 */ 75 class CLDirectDeconvolutionLayer : public IFunction 76 { 77 public: 78 /** Constructor */ 79 CLDirectDeconvolutionLayer(std::shared_ptr<IMemoryManager> memory_manager = nullptr); 80 /** Prevent instances of this class from being copied (As this class contains pointers) */ 81 CLDirectDeconvolutionLayer(const CLDirectDeconvolutionLayer &) = delete; 82 /** Default move constructor */ 83 CLDirectDeconvolutionLayer(CLDirectDeconvolutionLayer &&) = default; 84 /** Prevent instances of this class from being copied (As this class contains pointers) */ 85 CLDirectDeconvolutionLayer &operator=(const CLDirectDeconvolutionLayer &) = delete; 86 /** Default move assignment operator */ 87 CLDirectDeconvolutionLayer &operator=(CLDirectDeconvolutionLayer &&) = default; 88 /** Set the input, weights, biases and output tensors. 89 * 90 * Valid data layouts: 91 * - NHWC 92 * - NCHW 93 * 94 * Valid data type configurations: 95 * |src0 |src1 |src2 |dst | 96 * |:--------------|:------------------|:------|:--------------| 97 * |F16 |F16 |F16 |F16 | 98 * |F32 |F32 |F32 |F32 | 99 * |QASYMM8 |QASYMM8 |S32 |QASYMM8 | 100 * |QASYMM8_SIGNED |QASYMM8_SIGNED |S32 |QASYMM8_SIGNED | 101 * |QASYMM8 |QSYMM8_PER_CHANNEL |S32 |QASYMM8 | 102 * |QASYMM8_SIGNED |QSYMM8_PER_CHANNEL |S32 |QASYMM8_SIGNED | 103 * 104 * @param[in,out] input Input tensor. 3 lower dimensions represent a single input, and an optional 4th dimension for batch of inputs. 105 * Data types supported: QASYMM8_SIGNED/QASYMM8/F16/F32. 106 * @param[in] weights The 4d weights with dimensions [width, height, IFM, OFM]. Data type supported: Same as @p input or QSYMM8_PER_CHANNEL if @p input is QASYMM8/QASYMM8_SIGNED. 107 * @param[in] bias (Optional) The biases have one dimension. 108 * Data type supported: Should match @p input data type, except for input of QASYMM8 and QASYMM8_SIGNED type where biases should be of S32 type 109 * @param[out] output Output tensor. The output has the same number of dimensions as the @p input. 110 * @param[in] info Contains padding and policies to be used in the deconvolution, this is decribed in @ref PadStrideInfo. 111 * @param[in] weights_info (Optional) Weights information needed for @ref CLConvolutionLayer, specifies if the weights tensor has been reshaped with @ref opencl::kernels::ClWeightsReshapeKernel. 112 * 113 */ 114 void configure(ICLTensor *input, ICLTensor *weights, const ICLTensor *bias, ICLTensor *output, const PadStrideInfo &info, const WeightsInfo &weights_info = WeightsInfo()); 115 /** Set the input, weights, biases and output tensors. 116 * 117 * @param[in] compile_context The compile context to be used. 118 * @param[in,out] input Input tensor. 3 lower dimensions represent a single input, and an optional 4th dimension for batch of inputs. 119 * Data types supported: QASYMM8_SIGNED/QASYMM8/F16/F32. 120 * @param[in] weights The 4d weights with dimensions [width, height, IFM, OFM]. Data type supported: Same as @p input or QSYMM8_PER_CHANNEL if @p input is QASYMM8/QASYMM8_SIGNED. 121 * @param[in] bias (Optional) The biases have one dimension. 122 * Data type supported: Should match @p input data type, except for input of QASYMM8 and QASYMM8_SIGNED type where biases should be of S32 type 123 * @param[out] output Output tensor. The output has the same number of dimensions as the @p input. 124 * @param[in] info Contains padding and policies to be used in the deconvolution, this is decribed in @ref PadStrideInfo. 125 * @param[in] weights_info (Optional) Weights information needed for @ref CLConvolutionLayer, specifies if the weights tensor has been reshaped with @ref opencl::kernels::ClWeightsReshapeKernel. 126 * 127 */ 128 void configure(const CLCompileContext &compile_context, ICLTensor *input, ICLTensor *weights, const ICLTensor *bias, ICLTensor *output, const PadStrideInfo &info, 129 const WeightsInfo &weights_info = WeightsInfo()); 130 /** Static function to check if given info will lead to a valid configuration of @ref CLDirectDeconvolutionLayer 131 * 132 * @param[in] input Input tensor info. 3 lower dimensions represent a single input, and an optional 4th dimension for batch of inputs. 133 * Data types supported: QASYMM8_SIGNED/QASYMM8/F16/F32. 134 * @param[in] weights The 4d weights info with dimensions [width, height, IFM, OFM]. Data type supported: Same as @p input or QSYMM8_PER_CHANNEL if @p input is QASYMM8/QASYMM8_SIGNED. 135 * @param[in] bias (Optional) The biases have one dimension. 136 * Data type supported: Should match @p input data type, except for input of QASYMM8 and QASYMM8_SIGNED type where biases should be of S32 type 137 * @param[in] output Output tensor info. The output has the same number of dimensions as the @p input. 138 * @param[in] info Contains padding and policies to be used in the deconvolution, this is decribed in @ref PadStrideInfo. 139 * @param[in] weights_info (Optional) Weights information needed for @ref CLConvolutionLayer, specifies if the weights tensor has been reshaped with @ref opencl::kernels::ClWeightsReshapeKernel. 140 * 141 * @return a status 142 */ 143 static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *bias, ITensorInfo *output, const PadStrideInfo &info, 144 const WeightsInfo &weights_info = WeightsInfo()); 145 146 // Inherited methods overridden: 147 void run() override; 148 void prepare() override; 149 150 private: 151 MemoryGroup _memory_group; 152 CLDeconvolutionLayerUpsample _scale_f; 153 CLConvolutionLayer _conv_f; 154 CLReverse _flip_weights; 155 156 CLTensor _scaled_output; 157 ICLTensor *_original_weights; 158 CLTensor _weights_flipped; 159 CLTensor _flip_axis; 160 161 bool _is_prepared; 162 }; 163 } // namespace arm_compute 164 #endif /* ARM_COMPUTE_CLDECONVOLUTIONLAYER_H */ 165