From 2b5f0f2574551f59970bb9d710bafad2bc4bbd4a Mon Sep 17 00:00:00 2001 From: Michalis Spyrou Date: Wed, 10 Jan 2018 14:08:50 +0000 Subject: COMPMID-782 Port examples to the new format Change-Id: Ib178a97c080ff650094d02ee49e2a0aa22376dd0 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/115717 Reviewed-by: Anthony Barbier Tested-by: Jenkins --- examples/graph_mobilenet.cpp | 244 ++++++++++++++++++++++--------------------- 1 file changed, 125 insertions(+), 119 deletions(-) (limited to 'examples/graph_mobilenet.cpp') diff --git a/examples/graph_mobilenet.cpp b/examples/graph_mobilenet.cpp index c468093e56..193e5c336e 100644 --- a/examples/graph_mobilenet.cpp +++ b/examples/graph_mobilenet.cpp @@ -29,142 +29,148 @@ #include +using namespace arm_compute::utils; using namespace arm_compute::graph; using namespace arm_compute::graph_utils; -namespace -{ -BranchLayer get_dwsc_node(const std::string &data_path, std::string &¶m_path, - unsigned int conv_filt, - PadStrideInfo dwc_pad_stride_info, PadStrideInfo conv_pad_stride_info) -{ - std::string total_path = "/cnn_data/mobilenet_v1_model/" + param_path + "_"; - SubGraph sg; - sg << DepthwiseConvolutionLayer( - 3U, 3U, - get_weights_accessor(data_path, total_path + "depthwise_depthwise_weights.npy"), - std::unique_ptr(nullptr), - dwc_pad_stride_info, - true) - << BatchNormalizationLayer( - get_weights_accessor(data_path, total_path + "depthwise_BatchNorm_moving_mean.npy"), - get_weights_accessor(data_path, total_path + "depthwise_BatchNorm_moving_variance.npy"), - get_weights_accessor(data_path, total_path + "depthwise_BatchNorm_beta.npy"), - get_weights_accessor(data_path, total_path + "depthwise_BatchNorm_gamma.npy"), - 0.001f) - << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::BOUNDED_RELU, 6.f)) - << ConvolutionLayer( - 1U, 1U, conv_filt, - get_weights_accessor(data_path, total_path + "pointwise_weights.npy"), - std::unique_ptr(nullptr), - conv_pad_stride_info) - << BatchNormalizationLayer( - get_weights_accessor(data_path, total_path + "pointwise_BatchNorm_moving_mean.npy"), - get_weights_accessor(data_path, total_path + "pointwise_BatchNorm_moving_variance.npy"), - get_weights_accessor(data_path, total_path + "pointwise_BatchNorm_beta.npy"), - get_weights_accessor(data_path, total_path + "pointwise_BatchNorm_gamma.npy"), - 0.001f) - << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::BOUNDED_RELU, 6.f)); - - return BranchLayer(std::move(sg)); -} -} // namespace - /** Example demonstrating how to implement MobileNet's network using the Compute Library's graph API * * @param[in] argc Number of arguments * @param[in] argv Arguments ( [optional] Target (0 = NEON, 1 = OpenCL), [optional] Path to the weights folder, [optional] image, [optional] labels ) */ -void main_graph_mobilenet(int argc, char **argv) +class GraphMobilenetExample : public Example { - std::string data_path; /* Path to the trainable data */ - std::string image; /* Image data */ - std::string label; /* Label data */ +public: + void do_setup(int argc, char **argv) override + { + std::string data_path; /* Path to the trainable data */ + std::string image; /* Image data */ + std::string label; /* Label data */ - constexpr float mean_r = 122.68f; /* Mean value to subtract from red channel */ - constexpr float mean_g = 116.67f; /* Mean value to subtract from green channel */ - constexpr float mean_b = 104.01f; /* Mean value to subtract from blue channel */ + constexpr float mean_r = 122.68f; /* Mean value to subtract from red channel */ + constexpr float mean_g = 116.67f; /* Mean value to subtract from green channel */ + constexpr float mean_b = 104.01f; /* Mean value to subtract from blue channel */ - // Set target. 0 (NEON), 1 (OpenCL). By default it is NEON - TargetHint target_hint = set_target_hint(argc > 1 ? std::strtol(argv[1], nullptr, 10) : 0); - ConvolutionMethodHint convolution_hint = target_hint == TargetHint::NEON ? ConvolutionMethodHint::GEMM : ConvolutionMethodHint::DIRECT; + // Set target. 0 (NEON), 1 (OpenCL). By default it is NEON + TargetHint target_hint = set_target_hint(argc > 1 ? std::strtol(argv[1], nullptr, 10) : 0); + ConvolutionMethodHint convolution_hint = target_hint == TargetHint::NEON ? ConvolutionMethodHint::GEMM : ConvolutionMethodHint::DIRECT; - // Parse arguments - if(argc < 2) - { - // Print help - std::cout << "Usage: " << argv[0] << " [target] [path_to_data] [image] [labels]\n\n"; - std::cout << "No data folder provided: using random values\n\n"; - } - else if(argc == 2) - { - std::cout << "Usage: " << argv[0] << " " << argv[1] << " [path_to_data] [image] [labels]\n\n"; - std::cout << "No data folder provided: using random values\n\n"; - } - else if(argc == 3) - { - data_path = argv[2]; - std::cout << "Usage: " << argv[0] << " " << argv[1] << " " << argv[2] << " [image] [labels]\n\n"; - std::cout << "No image provided: using random values\n\n"; - } - else if(argc == 4) - { - data_path = argv[2]; - image = argv[3]; - std::cout << "Usage: " << argv[0] << " " << argv[1] << " " << argv[2] << " " << argv[3] << " [labels]\n\n"; - std::cout << "No text file with labels provided: skipping output accessor\n\n"; + // Parse arguments + if(argc < 2) + { + // Print help + std::cout << "Usage: " << argv[0] << " [target] [path_to_data] [image] [labels]\n\n"; + std::cout << "No data folder provided: using random values\n\n"; + } + else if(argc == 2) + { + std::cout << "Usage: " << argv[0] << " " << argv[1] << " [path_to_data] [image] [labels]\n\n"; + std::cout << "No data folder provided: using random values\n\n"; + } + else if(argc == 3) + { + data_path = argv[2]; + std::cout << "Usage: " << argv[0] << " " << argv[1] << " " << argv[2] << " [image] [labels]\n\n"; + std::cout << "No image provided: using random values\n\n"; + } + else if(argc == 4) + { + data_path = argv[2]; + image = argv[3]; + std::cout << "Usage: " << argv[0] << " " << argv[1] << " " << argv[2] << " " << argv[3] << " [labels]\n\n"; + std::cout << "No text file with labels provided: skipping output accessor\n\n"; + } + else + { + data_path = argv[2]; + image = argv[3]; + label = argv[4]; + } + + graph << target_hint + << Tensor(TensorInfo(TensorShape(224U, 224U, 3U, 1U), 1, DataType::F32), + get_input_accessor(image, mean_r, mean_g, mean_b)) + << convolution_hint + << ConvolutionLayer( + 3U, 3U, 32U, + get_weights_accessor(data_path, "/cnn_data/mobilenet_v1_model/Conv2d_0_weights.npy"), + std::unique_ptr(nullptr), + PadStrideInfo(2, 2, 0, 1, 0, 1, DimensionRoundingType::FLOOR)) + << BatchNormalizationLayer( + get_weights_accessor(data_path, "/cnn_data/mobilenet_v1_model/Conv2d_0_BatchNorm_moving_mean.npy"), + get_weights_accessor(data_path, "/cnn_data/mobilenet_v1_model/Conv2d_0_BatchNorm_moving_variance.npy"), + get_weights_accessor(data_path, "/cnn_data/mobilenet_v1_model/Conv2d_0_BatchNorm_beta.npy"), + get_weights_accessor(data_path, "/cnn_data/mobilenet_v1_model/Conv2d_0_BatchNorm_gamma.npy"), + 0.001f) + + << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::BOUNDED_RELU, 6.f)) + << get_dwsc_node(data_path, "Conv2d_1", 64, PadStrideInfo(1, 1, 1, 1), PadStrideInfo(1, 1, 0, 0)) + << get_dwsc_node(data_path, "Conv2d_2", 128, PadStrideInfo(2, 2, 0, 1, 0, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) + << get_dwsc_node(data_path, "Conv2d_3", 128, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) + << get_dwsc_node(data_path, "Conv2d_4", 256, PadStrideInfo(2, 2, 0, 1, 0, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) + << get_dwsc_node(data_path, "Conv2d_5", 256, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) + << get_dwsc_node(data_path, "Conv2d_6", 512, PadStrideInfo(2, 2, 0, 1, 0, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) + << get_dwsc_node(data_path, "Conv2d_7", 512, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) + << get_dwsc_node(data_path, "Conv2d_8", 512, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) + << get_dwsc_node(data_path, "Conv2d_9", 512, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) + << get_dwsc_node(data_path, "Conv2d_10", 512, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) + << get_dwsc_node(data_path, "Conv2d_11", 512, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) + << get_dwsc_node(data_path, "Conv2d_12", 1024, PadStrideInfo(2, 2, 0, 1, 0, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) + << get_dwsc_node(data_path, "Conv2d_13", 1024, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) + << PoolingLayer(PoolingLayerInfo(PoolingType::AVG)) + << ConvolutionLayer( + 1U, 1U, 1001U, + get_weights_accessor(data_path, "/cnn_data/mobilenet_v1_model/Logits_Conv2d_1c_1x1_weights.npy"), + get_weights_accessor(data_path, "/cnn_data/mobilenet_v1_model/Logits_Conv2d_1c_1x1_biases.npy"), + PadStrideInfo(1, 1, 0, 0)) + << ReshapeLayer(TensorShape(1001U)) + << SoftmaxLayer() + << Tensor(get_output_accessor(label, 5)); } - else + void do_run() override { - data_path = argv[2]; - image = argv[3]; - label = argv[4]; + // Run graph + graph.run(); } - Graph graph; +private: + Graph graph{}; - graph << target_hint - << Tensor(TensorInfo(TensorShape(224U, 224U, 3U, 1U), 1, DataType::F32), - get_input_accessor(image, mean_r, mean_g, mean_b)) - << convolution_hint - << ConvolutionLayer( - 3U, 3U, 32U, - get_weights_accessor(data_path, "/cnn_data/mobilenet_v1_model/Conv2d_0_weights.npy"), - std::unique_ptr(nullptr), - PadStrideInfo(2, 2, 0, 1, 0, 1, DimensionRoundingType::FLOOR)) - << BatchNormalizationLayer( - get_weights_accessor(data_path, "/cnn_data/mobilenet_v1_model/Conv2d_0_BatchNorm_moving_mean.npy"), - get_weights_accessor(data_path, "/cnn_data/mobilenet_v1_model/Conv2d_0_BatchNorm_moving_variance.npy"), - get_weights_accessor(data_path, "/cnn_data/mobilenet_v1_model/Conv2d_0_BatchNorm_beta.npy"), - get_weights_accessor(data_path, "/cnn_data/mobilenet_v1_model/Conv2d_0_BatchNorm_gamma.npy"), - 0.001f) - - << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::BOUNDED_RELU, 6.f)) - << get_dwsc_node(data_path, "Conv2d_1", 64, PadStrideInfo(1, 1, 1, 1), PadStrideInfo(1, 1, 0, 0)) - << get_dwsc_node(data_path, "Conv2d_2", 128, PadStrideInfo(2, 2, 0, 1, 0, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) - << get_dwsc_node(data_path, "Conv2d_3", 128, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) - << get_dwsc_node(data_path, "Conv2d_4", 256, PadStrideInfo(2, 2, 0, 1, 0, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) - << get_dwsc_node(data_path, "Conv2d_5", 256, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) - << get_dwsc_node(data_path, "Conv2d_6", 512, PadStrideInfo(2, 2, 0, 1, 0, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) - << get_dwsc_node(data_path, "Conv2d_7", 512, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) - << get_dwsc_node(data_path, "Conv2d_8", 512, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) - << get_dwsc_node(data_path, "Conv2d_9", 512, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) - << get_dwsc_node(data_path, "Conv2d_10", 512, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) - << get_dwsc_node(data_path, "Conv2d_11", 512, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) - << get_dwsc_node(data_path, "Conv2d_12", 1024, PadStrideInfo(2, 2, 0, 1, 0, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) - << get_dwsc_node(data_path, "Conv2d_13", 1024, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::FLOOR), PadStrideInfo(1, 1, 0, 0)) - << PoolingLayer(PoolingLayerInfo(PoolingType::AVG)) - << ConvolutionLayer( - 1U, 1U, 1001U, - get_weights_accessor(data_path, "/cnn_data/mobilenet_v1_model/Logits_Conv2d_1c_1x1_weights.npy"), - get_weights_accessor(data_path, "/cnn_data/mobilenet_v1_model/Logits_Conv2d_1c_1x1_biases.npy"), - PadStrideInfo(1, 1, 0, 0)) - << ReshapeLayer(TensorShape(1001U)) - << SoftmaxLayer() - << Tensor(get_output_accessor(label, 5)); + BranchLayer get_dwsc_node(const std::string &data_path, std::string &¶m_path, + unsigned int conv_filt, + PadStrideInfo dwc_pad_stride_info, PadStrideInfo conv_pad_stride_info) + { + std::string total_path = "/cnn_data/mobilenet_v1_model/" + param_path + "_"; + SubGraph sg; + sg << DepthwiseConvolutionLayer( + 3U, 3U, + get_weights_accessor(data_path, total_path + "depthwise_depthwise_weights.npy"), + std::unique_ptr(nullptr), + dwc_pad_stride_info, + true) + << BatchNormalizationLayer( + get_weights_accessor(data_path, total_path + "depthwise_BatchNorm_moving_mean.npy"), + get_weights_accessor(data_path, total_path + "depthwise_BatchNorm_moving_variance.npy"), + get_weights_accessor(data_path, total_path + "depthwise_BatchNorm_beta.npy"), + get_weights_accessor(data_path, total_path + "depthwise_BatchNorm_gamma.npy"), + 0.001f) + << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::BOUNDED_RELU, 6.f)) + << ConvolutionLayer( + 1U, 1U, conv_filt, + get_weights_accessor(data_path, total_path + "pointwise_weights.npy"), + std::unique_ptr(nullptr), + conv_pad_stride_info) + << BatchNormalizationLayer( + get_weights_accessor(data_path, total_path + "pointwise_BatchNorm_moving_mean.npy"), + get_weights_accessor(data_path, total_path + "pointwise_BatchNorm_moving_variance.npy"), + get_weights_accessor(data_path, total_path + "pointwise_BatchNorm_beta.npy"), + get_weights_accessor(data_path, total_path + "pointwise_BatchNorm_gamma.npy"), + 0.001f) + << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::BOUNDED_RELU, 6.f)); - graph.run(); -} + return BranchLayer(std::move(sg)); + } +}; /** Main program for MobileNetV1 * @@ -173,5 +179,5 @@ void main_graph_mobilenet(int argc, char **argv) */ int main(int argc, char **argv) { - return arm_compute::utils::run_example(argc, argv, main_graph_mobilenet); + return arm_compute::utils::run_example(argc, argv); } -- cgit v1.2.1