Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / ComputeLibrary / 292227986edb37b01061afcad6df18ba9d6ccbeb / . / examples / graph_vgg16.cpp
blob: 1a804a4882c72b8811eb9d794f0d74c045583be4 [file] [log] [blame]
Anthony Barbier8140e1e2017-12-14 23:48:46 +0000[diff] [blame]1/*
Anthony Barbierf45d5a92018-01-24 16:23:15 +0000[diff] [blame]2 * Copyright (c) 2017, 2018 ARM Limited.
Anthony Barbier8140e1e2017-12-14 23:48:46 +0000[diff] [blame]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#include "arm_compute/graph/Graph.h"
25#include "arm_compute/graph/Nodes.h"
26#include "support/ToolchainSupport.h"
27#include "utils/GraphUtils.h"
28#include "utils/Utils.h"
29
30#include <cstdlib>
31
Anthony Barbierf45d5a92018-01-24 16:23:15 +0000[diff] [blame]32using namespace arm_compute::utils;
Anthony Barbier8140e1e2017-12-14 23:48:46 +0000[diff] [blame]33using namespace arm_compute::graph;
34using namespace arm_compute::graph_utils;
35
36/** Example demonstrating how to implement VGG16's network using the Compute Library's graph API
37 *
38 * @param[in] argc Number of arguments
39 * @param[in] argv Arguments ( [optional] Target (0 = NEON, 1 = OpenCL), [optional] Path to the weights folder, [optional] image, [optional] labels )
40 */
Anthony Barbierf45d5a92018-01-24 16:23:15 +0000[diff] [blame]41class GraphVGG16Example : public Example
Anthony Barbier8140e1e2017-12-14 23:48:46 +0000[diff] [blame]42{
Anthony Barbierf45d5a92018-01-24 16:23:15 +0000[diff] [blame]43public:
44 void do_setup(int argc, char **argv) override
45 {
46 std::string data_path; /* Path to the trainable data */
47 std::string image; /* Image data */
48 std::string label; /* Label data */
Anthony Barbier8140e1e2017-12-14 23:48:46 +0000[diff] [blame]49
Anthony Barbierf45d5a92018-01-24 16:23:15 +0000[diff] [blame]50 constexpr float mean_r = 123.68f; /* Mean value to subtract from red channel */
51 constexpr float mean_g = 116.779f; /* Mean value to subtract from green channel */
52 constexpr float mean_b = 103.939f; /* Mean value to subtract from blue channel */
Anthony Barbier8140e1e2017-12-14 23:48:46 +0000[diff] [blame]53
Anthony Barbierf45d5a92018-01-24 16:23:15 +0000[diff] [blame]54 // Set target. 0 (NEON), 1 (OpenCL). By default it is NEON
55 TargetHint target_hint = set_target_hint(argc > 1 ? std::strtol(argv[1], nullptr, 10) : 0);
56 ConvolutionMethodHint convolution_hint = ConvolutionMethodHint::DIRECT;
Anthony Barbier8140e1e2017-12-14 23:48:46 +0000[diff] [blame]57
Anthony Barbierf45d5a92018-01-24 16:23:15 +0000[diff] [blame]58 // Parse arguments
59 if(argc < 2)
60 {
61 // Print help
62 std::cout << "Usage: " << argv[0] << " [target] [path_to_data] [image] [labels]\n\n";
63 std::cout << "No data folder provided: using random values\n\n";
64 }
65 else if(argc == 2)
66 {
67 std::cout << "Usage: " << argv[0] << " " << argv[1] << " [path_to_data] [image] [labels]\n\n";
68 std::cout << "No data folder provided: using random values\n\n";
69 }
70 else if(argc == 3)
71 {
72 data_path = argv[2];
73 std::cout << "Usage: " << argv[0] << " " << argv[1] << " " << argv[2] << " [image] [labels]\n\n";
74 std::cout << "No image provided: using random values\n\n";
75 }
76 else if(argc == 4)
77 {
78 data_path = argv[2];
79 image = argv[3];
80 std::cout << "Usage: " << argv[0] << " " << argv[1] << " " << argv[2] << " " << argv[3] << " [labels]\n\n";
81 std::cout << "No text file with labels provided: skipping output accessor\n\n";
82 }
83 else
84 {
85 data_path = argv[2];
86 image = argv[3];
87 label = argv[4];
88 }
89
90 graph << target_hint
91 << convolution_hint
92 << Tensor(TensorInfo(TensorShape(224U, 224U, 3U, 1U), 1, DataType::F32),
93 get_input_accessor(image, mean_r, mean_g, mean_b))
94 << ConvolutionMethodHint::DIRECT
95 // Layer 1
96 << ConvolutionLayer(
97 3U, 3U, 64U,
98 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv1_1_w.npy"),
99 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv1_1_b.npy"),
100 PadStrideInfo(1, 1, 1, 1))
101 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
102 // Layer 2
103 << ConvolutionLayer(
104 3U, 3U, 64U,
105 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv1_2_w.npy"),
106 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv1_2_b.npy"),
107 PadStrideInfo(1, 1, 1, 1))
108 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
109 << PoolingLayer(PoolingLayerInfo(PoolingType::MAX, 2, PadStrideInfo(2, 2, 0, 0)))
110 // Layer 3
111 << ConvolutionLayer(
112 3U, 3U, 128U,
113 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv2_1_w.npy"),
114 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv2_1_b.npy"),
115 PadStrideInfo(1, 1, 1, 1))
116 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
117 // Layer 4
118 << ConvolutionLayer(
119 3U, 3U, 128U,
120 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv2_2_w.npy"),
121 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv2_2_b.npy"),
122 PadStrideInfo(1, 1, 1, 1))
123 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
124 << PoolingLayer(PoolingLayerInfo(PoolingType::MAX, 2, PadStrideInfo(2, 2, 0, 0)))
125 // Layer 5
126 << ConvolutionLayer(
127 3U, 3U, 256U,
128 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv3_1_w.npy"),
129 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv3_1_b.npy"),
130 PadStrideInfo(1, 1, 1, 1))
131 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
132 // Layer 6
133 << ConvolutionLayer(
134 3U, 3U, 256U,
135 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv3_2_w.npy"),
136 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv3_2_b.npy"),
137 PadStrideInfo(1, 1, 1, 1))
138 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
139 // Layer 7
140 << ConvolutionLayer(
141 3U, 3U, 256U,
142 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv3_3_w.npy"),
143 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv3_3_b.npy"),
144 PadStrideInfo(1, 1, 1, 1))
145 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
146 << PoolingLayer(PoolingLayerInfo(PoolingType::MAX, 2, PadStrideInfo(2, 2, 0, 0)))
147 // Layer 8
148 << ConvolutionLayer(
149 3U, 3U, 512U,
150 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv4_1_w.npy"),
151 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv4_1_b.npy"),
152 PadStrideInfo(1, 1, 1, 1))
153 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
154 // Layer 9
155 << ConvolutionLayer(
156 3U, 3U, 512U,
157 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv4_2_w.npy"),
158 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv4_2_b.npy"),
159 PadStrideInfo(1, 1, 1, 1))
160 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
161 // Layer 10
162 << ConvolutionLayer(
163 3U, 3U, 512U,
164 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv4_3_w.npy"),
165 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv4_3_b.npy"),
166 PadStrideInfo(1, 1, 1, 1))
167 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
168 << PoolingLayer(PoolingLayerInfo(PoolingType::MAX, 2, PadStrideInfo(2, 2, 0, 0)))
169 // Layer 11
170 << ConvolutionLayer(
171 3U, 3U, 512U,
172 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv5_1_w.npy"),
173 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv5_1_b.npy"),
174 PadStrideInfo(1, 1, 1, 1))
175 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
176 // Layer 12
177 << ConvolutionLayer(
178 3U, 3U, 512U,
179 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv5_2_w.npy"),
180 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv5_2_b.npy"),
181 PadStrideInfo(1, 1, 1, 1))
182 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
183 // Layer 13
184 << ConvolutionLayer(
185 3U, 3U, 512U,
186 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv5_3_w.npy"),
187 get_weights_accessor(data_path, "/cnn_data/vgg16_model/conv5_3_b.npy"),
188 PadStrideInfo(1, 1, 1, 1))
189 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
190 << PoolingLayer(PoolingLayerInfo(PoolingType::MAX, 2, PadStrideInfo(2, 2, 0, 0)))
191 // Layer 14
192 << FullyConnectedLayer(
193 4096U,
194 get_weights_accessor(data_path, "/cnn_data/vgg16_model/fc6_w.npy"),
195 get_weights_accessor(data_path, "/cnn_data/vgg16_model/fc6_b.npy"))
196 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
197 // Layer 15
198 << FullyConnectedLayer(
199 4096U,
200 get_weights_accessor(data_path, "/cnn_data/vgg16_model/fc7_w.npy"),
201 get_weights_accessor(data_path, "/cnn_data/vgg16_model/fc7_b.npy"))
202 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
203 // Layer 16
204 << FullyConnectedLayer(
205 1000U,
206 get_weights_accessor(data_path, "/cnn_data/vgg16_model/fc8_w.npy"),
207 get_weights_accessor(data_path, "/cnn_data/vgg16_model/fc8_b.npy"))
208 // Softmax
209 << SoftmaxLayer()
210 << Tensor(get_output_accessor(label, 5));
Anthony Barbier8140e1e2017-12-14 23:48:46 +0000[diff] [blame]211 }
Anthony Barbierf45d5a92018-01-24 16:23:15 +0000[diff] [blame]212 void do_run() override
Anthony Barbier8140e1e2017-12-14 23:48:46 +0000[diff] [blame]213 {
Anthony Barbierf45d5a92018-01-24 16:23:15 +0000[diff] [blame]214 // Run graph
215 graph.run();
Anthony Barbier8140e1e2017-12-14 23:48:46 +0000[diff] [blame]216 }
217
Anthony Barbierf45d5a92018-01-24 16:23:15 +0000[diff] [blame]218private:
219 Graph graph{};
220};
Anthony Barbier8140e1e2017-12-14 23:48:46 +0000[diff] [blame]221
222/** Main program for VGG16
223 *
224 * @param[in] argc Number of arguments
225 * @param[in] argv Arguments ( [optional] Target (0 = NEON, 1 = OpenCL), [optional] Path to the weights folder, [optional] image, [optional] labels )
226 */
Anthony Barbierf45d5a92018-01-24 16:23:15 +0000[diff] [blame]227int main(int argc, char **argv)
Anthony Barbier8140e1e2017-12-14 23:48:46 +0000[diff] [blame]228{
Anthony Barbierf45d5a92018-01-24 16:23:15 +0000[diff] [blame]229 return arm_compute::utils::run_example<GraphVGG16Example>(argc, argv);
Anthony Barbier8140e1e2017-12-14 23:48:46 +0000[diff] [blame]230}