1 /*
2 * Copyright (c) 2017-2021 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 #include "arm_compute/graph.h"
25 #include "support/ToolchainSupport.h"
26 #include "utils/CommonGraphOptions.h"
27 #include "utils/GraphUtils.h"
28 #include "utils/Utils.h"
29
30 using namespace arm_compute;
31 using namespace arm_compute::utils;
32 using namespace arm_compute::graph::frontend;
33 using namespace arm_compute::graph_utils;
34
35 /** Example demonstrating how to implement MobileNet's network using the Compute Library's graph API */
36 class GraphMobilenetExample : public Example
37 {
38 public:
GraphMobilenetExample()39 GraphMobilenetExample()
40 : cmd_parser(), common_opts(cmd_parser), common_params(), graph(0, "MobileNetV1")
41 {
42 // Add model id option
43 model_id_opt = cmd_parser.add_option<SimpleOption<int>>("model-id", 0);
44 model_id_opt->set_help("Mobilenet model id (0: 1.0_224, else: 0.75_160");
45 }
46 GraphMobilenetExample(const GraphMobilenetExample &) = delete;
47 GraphMobilenetExample &operator=(const GraphMobilenetExample &) = delete;
48 ~GraphMobilenetExample() override = default;
do_setup(int argc,char ** argv)49 bool do_setup(int argc, char **argv) override
50 {
51 // Parse arguments
52 cmd_parser.parse(argc, argv);
53 cmd_parser.validate();
54
55 // Consume common parameters
56 common_params = consume_common_graph_parameters(common_opts);
57
58 // Return when help menu is requested
59 if(common_params.help)
60 {
61 cmd_parser.print_help(argv[0]);
62 return false;
63 }
64
65 // Print parameter values
66 std::cout << common_params << std::endl;
67
68 // Get model parameters
69 int model_id = model_id_opt->value();
70
71 // Create input descriptor
72 unsigned int spatial_size = (model_id == 0 || common_params.data_type == DataType::QASYMM8) ? 224 : 160;
73
74 // Create input descriptor
75 const TensorShape tensor_shape = permute_shape(TensorShape(spatial_size, spatial_size, 3U, common_params.batches), DataLayout::NCHW, common_params.data_layout);
76 TensorDescriptor input_descriptor = TensorDescriptor(tensor_shape, common_params.data_type).set_layout(common_params.data_layout);
77
78 // Set graph hints
79 graph << common_params.target
80 << common_params.fast_math_hint;
81
82 // Create core graph
83 if(arm_compute::is_data_type_float(common_params.data_type))
84 {
85 create_graph_float(input_descriptor, model_id);
86 }
87 else
88 {
89 create_graph_qasymm(input_descriptor);
90 }
91
92 // Create common tail
93 graph << ReshapeLayer(TensorShape(1001U)).set_name("Reshape")
94 << SoftmaxLayer().set_name("Softmax")
95 << OutputLayer(get_output_accessor(common_params, 5));
96
97 // Finalize graph
98 GraphConfig config;
99 config.num_threads = common_params.threads;
100 config.use_tuner = common_params.enable_tuner;
101 config.tuner_mode = common_params.tuner_mode;
102 config.tuner_file = common_params.tuner_file;
103 config.mlgo_file = common_params.mlgo_file;
104
105 graph.finalize(common_params.target, config);
106
107 return true;
108 }
do_run()109 void do_run() override
110 {
111 // Run graph
112 graph.run();
113 }
114
115 private:
116 CommandLineParser cmd_parser;
117 CommonGraphOptions common_opts;
118 SimpleOption<int> *model_id_opt{ nullptr };
119 CommonGraphParams common_params;
120 Stream graph;
121
create_graph_float(TensorDescriptor & input_descriptor,int model_id)122 void create_graph_float(TensorDescriptor &input_descriptor, int model_id)
123 {
124 float depth_scale = (model_id == 0) ? 1.f : 0.75;
125 std::string model_path = (model_id == 0) ? "/cnn_data/mobilenet_v1_1_224_model/" : "/cnn_data/mobilenet_v1_075_160_model/";
126
127 // Create a preprocessor object
128 std::unique_ptr<IPreprocessor> preprocessor = std::make_unique<TFPreproccessor>();
129
130 // Get trainable parameters data path
131 std::string data_path = common_params.data_path;
132
133 // Add model path to data path
134 if(!data_path.empty())
135 {
136 data_path += model_path;
137 }
138
139 graph << InputLayer(input_descriptor,
140 get_input_accessor(common_params, std::move(preprocessor), false))
141 << ConvolutionLayer(
142 3U, 3U, 32U * depth_scale,
143 get_weights_accessor(data_path, "Conv2d_0_weights.npy", DataLayout::NCHW),
144 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
145 PadStrideInfo(2, 2, 0, 1, 0, 1, DimensionRoundingType::FLOOR))
146 .set_name("Conv2d_0")
147 << BatchNormalizationLayer(
148 get_weights_accessor(data_path, "Conv2d_0_BatchNorm_moving_mean.npy"),
149 get_weights_accessor(data_path, "Conv2d_0_BatchNorm_moving_variance.npy"),
150 get_weights_accessor(data_path, "Conv2d_0_BatchNorm_gamma.npy"),
151 get_weights_accessor(data_path, "Conv2d_0_BatchNorm_beta.npy"),
152 0.001f)
153 .set_name("Conv2d_0/BatchNorm")
154 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::BOUNDED_RELU, 6.f)).set_name("Conv2d_0/Relu6");
155 graph << get_dwsc_node_float(data_path, "Conv2d_1", 64 * depth_scale, PadStrideInfo(1, 1, 1, 1), PadStrideInfo(1, 1, 0, 0));
156 graph << get_dwsc_node_float(data_path, "Conv2d_2", 128 * depth_scale, PadStrideInfo(2, 2, 0, 1, 0, 1, DimensionRoundingType::CEIL), PadStrideInfo(1, 1, 0, 0));
157 graph << get_dwsc_node_float(data_path, "Conv2d_3", 128 * depth_scale, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::CEIL), PadStrideInfo(1, 1, 0, 0));
158 graph << get_dwsc_node_float(data_path, "Conv2d_4", 256 * depth_scale, PadStrideInfo(2, 2, 0, 1, 0, 1, DimensionRoundingType::CEIL), PadStrideInfo(1, 1, 0, 0));
159 graph << get_dwsc_node_float(data_path, "Conv2d_5", 256 * depth_scale, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::CEIL), PadStrideInfo(1, 1, 0, 0));
160 graph << get_dwsc_node_float(data_path, "Conv2d_6", 512 * depth_scale, PadStrideInfo(2, 2, 0, 1, 0, 1, DimensionRoundingType::CEIL), PadStrideInfo(1, 1, 0, 0));
161 graph << get_dwsc_node_float(data_path, "Conv2d_7", 512 * depth_scale, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::CEIL), PadStrideInfo(1, 1, 0, 0));
162 graph << get_dwsc_node_float(data_path, "Conv2d_8", 512 * depth_scale, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::CEIL), PadStrideInfo(1, 1, 0, 0));
163 graph << get_dwsc_node_float(data_path, "Conv2d_9", 512 * depth_scale, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::CEIL), PadStrideInfo(1, 1, 0, 0));
164 graph << get_dwsc_node_float(data_path, "Conv2d_10", 512 * depth_scale, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::CEIL), PadStrideInfo(1, 1, 0, 0));
165 graph << get_dwsc_node_float(data_path, "Conv2d_11", 512 * depth_scale, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::CEIL), PadStrideInfo(1, 1, 0, 0));
166 graph << get_dwsc_node_float(data_path, "Conv2d_12", 1024 * depth_scale, PadStrideInfo(2, 2, 0, 1, 0, 1, DimensionRoundingType::CEIL), PadStrideInfo(1, 1, 0, 0));
167 graph << get_dwsc_node_float(data_path, "Conv2d_13", 1024 * depth_scale, PadStrideInfo(1, 1, 1, 1, 1, 1, DimensionRoundingType::CEIL), PadStrideInfo(1, 1, 0, 0));
168 graph << PoolingLayer(PoolingLayerInfo(PoolingType::AVG, common_params.data_layout)).set_name("Logits/AvgPool_1a")
169 << ConvolutionLayer(
170 1U, 1U, 1001U,
171 get_weights_accessor(data_path, "Logits_Conv2d_1c_1x1_weights.npy", DataLayout::NCHW),
172 get_weights_accessor(data_path, "Logits_Conv2d_1c_1x1_biases.npy"),
173 PadStrideInfo(1, 1, 0, 0))
174 .set_name("Logits/Conv2d_1c_1x1");
175 }
176
create_graph_qasymm(TensorDescriptor & input_descriptor)177 void create_graph_qasymm(TensorDescriptor &input_descriptor)
178 {
179 // Get trainable parameters data path
180 std::string data_path = common_params.data_path;
181
182 // Add model path to data path
183 if(!data_path.empty())
184 {
185 data_path += "/cnn_data/mobilenet_qasymm8_model/";
186 }
187
188 // Quantization info taken from the AndroidNN QASYMM8 MobileNet example
189 const QuantizationInfo in_quant_info = QuantizationInfo(0.0078125f, 128);
190
191 const std::vector<QuantizationInfo> conv_weights_quant_info =
192 {
193 QuantizationInfo(0.02182667888700962f, 151), // conv0
194 QuantizationInfo(0.004986600950360298f, 74) // conv14
195 };
196 const std::vector<QuantizationInfo> conv_out_quant_info =
197 {
198 QuantizationInfo(0.023528477177023888f, 0), // conv0
199 QuantizationInfo(0.16609922051429749f, 66) // conv14
200 };
201
202 const std::vector<QuantizationInfo> depth_weights_quant_info =
203 {
204 QuantizationInfo(0.29219913482666016f, 110), // dwsc1
205 QuantizationInfo(0.40277284383773804f, 130), // dwsc2
206 QuantizationInfo(0.06053730100393295f, 160), // dwsc3
207 QuantizationInfo(0.01675807684659958f, 123), // dwsc4
208 QuantizationInfo(0.04105526953935623f, 129), // dwsc5
209 QuantizationInfo(0.013460792601108551f, 122), // dwsc6
210 QuantizationInfo(0.036934755742549896f, 132), // dwsc7
211 QuantizationInfo(0.042609862983226776f, 94), // dwsc8
212 QuantizationInfo(0.028358859941363335f, 127), // dwsc9
213 QuantizationInfo(0.024329448118805885f, 134), // dwsc10
214 QuantizationInfo(0.019366811960935593f, 106), // dwsc11
215 QuantizationInfo(0.007835594937205315f, 126), // dwsc12
216 QuantizationInfo(0.12616927921772003f, 211) // dwsc13
217 };
218
219 const std::vector<QuantizationInfo> point_weights_quant_info =
220 {
221 QuantizationInfo(0.030420949682593346f, 121), // dwsc1
222 QuantizationInfo(0.015148180536925793f, 104), // dwsc2
223 QuantizationInfo(0.013755458407104015f, 94), // dwsc3
224 QuantizationInfo(0.007601846940815449f, 151), // dwsc4
225 QuantizationInfo(0.006431614048779011f, 122), // dwsc5
226 QuantizationInfo(0.00917122047394514f, 109), // dwsc6
227 QuantizationInfo(0.005300046876072884f, 140), // dwsc7
228 QuantizationInfo(0.0049632852897048f, 127), // dwsc8
229 QuantizationInfo(0.007770895957946777f, 89), // dwsc9
230 QuantizationInfo(0.009658650495111942f, 99), // dwsc10
231 QuantizationInfo(0.005446993745863438f, 153), // dwsc11
232 QuantizationInfo(0.00817922968417406f, 130), // dwsc12
233 QuantizationInfo(0.018048152327537537f, 95) // dwsc13
234 };
235
236 graph << InputLayer(input_descriptor.set_quantization_info(in_quant_info),
237 get_input_accessor(common_params, nullptr, false))
238 << ConvolutionLayer(
239 3U, 3U, 32U,
240 get_weights_accessor(data_path, "Conv2d_0_weights.npy"),
241 get_weights_accessor(data_path, "Conv2d_0_bias.npy"),
242 PadStrideInfo(2U, 2U, 0U, 1U, 0U, 1U, DimensionRoundingType::FLOOR),
243 1, conv_weights_quant_info.at(0), conv_out_quant_info.at(0))
244 .set_name("Conv2d_0")
245 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::LU_BOUNDED_RELU, 6.f)).set_name("Conv2d_0/Relu6");
246 graph << get_dwsc_node_qasymm(data_path, "Conv2d_1", 64U, PadStrideInfo(1U, 1U, 1U, 1U), PadStrideInfo(1U, 1U, 0U, 0U), depth_weights_quant_info.at(0), point_weights_quant_info.at(0));
247 graph << get_dwsc_node_qasymm(data_path, "Conv2d_2", 128U, PadStrideInfo(2U, 2U, 0U, 1U, 0U, 1U, DimensionRoundingType::FLOOR), PadStrideInfo(1U, 1U, 0U, 0U), depth_weights_quant_info.at(1),
248 point_weights_quant_info.at(1));
249 graph << get_dwsc_node_qasymm(data_path, "Conv2d_3", 128U, PadStrideInfo(1U, 1U, 1U, 1U, 1U, 1U, DimensionRoundingType::FLOOR), PadStrideInfo(1U, 1U, 0U, 0U), depth_weights_quant_info.at(2),
250 point_weights_quant_info.at(2));
251 graph << get_dwsc_node_qasymm(data_path, "Conv2d_4", 256U, PadStrideInfo(2U, 2U, 0U, 1U, 0U, 1U, DimensionRoundingType::FLOOR), PadStrideInfo(1U, 1U, 0U, 0U), depth_weights_quant_info.at(3),
252 point_weights_quant_info.at(3));
253 graph << get_dwsc_node_qasymm(data_path, "Conv2d_5", 256U, PadStrideInfo(1U, 1U, 1U, 1U, 1U, 1U, DimensionRoundingType::FLOOR), PadStrideInfo(1U, 1U, 0U, 0U), depth_weights_quant_info.at(4),
254 point_weights_quant_info.at(4));
255 graph << get_dwsc_node_qasymm(data_path, "Conv2d_6", 512U, PadStrideInfo(2U, 2U, 0U, 1U, 0U, 1U, DimensionRoundingType::FLOOR), PadStrideInfo(1U, 1U, 0U, 0U), depth_weights_quant_info.at(5),
256 point_weights_quant_info.at(5));
257 graph << get_dwsc_node_qasymm(data_path, "Conv2d_7", 512U, PadStrideInfo(1U, 1U, 1U, 1U, 1U, 1U, DimensionRoundingType::FLOOR), PadStrideInfo(1U, 1U, 0U, 0U), depth_weights_quant_info.at(6),
258 point_weights_quant_info.at(6));
259 graph << get_dwsc_node_qasymm(data_path, "Conv2d_8", 512U, PadStrideInfo(1U, 1U, 1U, 1U, 1U, 1U, DimensionRoundingType::FLOOR), PadStrideInfo(1U, 1U, 0U, 0U), depth_weights_quant_info.at(7),
260 point_weights_quant_info.at(7));
261 graph << get_dwsc_node_qasymm(data_path, "Conv2d_9", 512U, PadStrideInfo(1U, 1U, 1U, 1U, 1U, 1U, DimensionRoundingType::FLOOR), PadStrideInfo(1U, 1U, 0U, 0U), depth_weights_quant_info.at(8),
262 point_weights_quant_info.at(8));
263 graph << get_dwsc_node_qasymm(data_path, "Conv2d_10", 512U, PadStrideInfo(1U, 1U, 1U, 1U, 1U, 1U, DimensionRoundingType::FLOOR), PadStrideInfo(1U, 1U, 0U, 0U), depth_weights_quant_info.at(9),
264 point_weights_quant_info.at(9));
265 graph << get_dwsc_node_qasymm(data_path, "Conv2d_11", 512U, PadStrideInfo(1U, 1U, 1U, 1U, 1U, 1U, DimensionRoundingType::FLOOR), PadStrideInfo(1U, 1U, 0U, 0U), depth_weights_quant_info.at(10),
266 point_weights_quant_info.at(10));
267 graph << get_dwsc_node_qasymm(data_path, "Conv2d_12", 1024U, PadStrideInfo(2U, 2U, 0U, 1U, 0U, 1U, DimensionRoundingType::FLOOR), PadStrideInfo(1U, 1U, 0U, 0U), depth_weights_quant_info.at(11),
268 point_weights_quant_info.at(11));
269 graph << get_dwsc_node_qasymm(data_path, "Conv2d_13", 1024U, PadStrideInfo(1U, 1U, 1U, 1U, 1U, 1U, DimensionRoundingType::FLOOR), PadStrideInfo(1U, 1U, 0U, 0U), depth_weights_quant_info.at(12),
270 point_weights_quant_info.at(12))
271 << PoolingLayer(PoolingLayerInfo(PoolingType::AVG, common_params.data_layout)).set_name("Logits/AvgPool_1a")
272 << ConvolutionLayer(
273 1U, 1U, 1001U,
274 get_weights_accessor(data_path, "Logits_Conv2d_1c_1x1_weights.npy"),
275 get_weights_accessor(data_path, "Logits_Conv2d_1c_1x1_bias.npy"),
276 PadStrideInfo(1U, 1U, 0U, 0U), 1, conv_weights_quant_info.at(1), conv_out_quant_info.at(1))
277 .set_name("Logits/Conv2d_1c_1x1");
278 }
279
get_dwsc_node_float(const std::string & data_path,std::string && param_path,unsigned int conv_filt,PadStrideInfo dwc_pad_stride_info,PadStrideInfo conv_pad_stride_info)280 ConcatLayer get_dwsc_node_float(const std::string &data_path, std::string &¶m_path,
281 unsigned int conv_filt,
282 PadStrideInfo dwc_pad_stride_info, PadStrideInfo conv_pad_stride_info)
283 {
284 std::string total_path = param_path + "_";
285 SubStream sg(graph);
286 sg << DepthwiseConvolutionLayer(
287 3U, 3U,
288 get_weights_accessor(data_path, total_path + "depthwise_depthwise_weights.npy", DataLayout::NCHW),
289 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
290 dwc_pad_stride_info)
291 .set_name(total_path + "depthwise/depthwise")
292 << BatchNormalizationLayer(
293 get_weights_accessor(data_path, total_path + "depthwise_BatchNorm_moving_mean.npy"),
294 get_weights_accessor(data_path, total_path + "depthwise_BatchNorm_moving_variance.npy"),
295 get_weights_accessor(data_path, total_path + "depthwise_BatchNorm_gamma.npy"),
296 get_weights_accessor(data_path, total_path + "depthwise_BatchNorm_beta.npy"),
297 0.001f)
298 .set_name(total_path + "depthwise/BatchNorm")
299 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::BOUNDED_RELU, 6.f)).set_name(total_path + "depthwise/Relu6")
300 << ConvolutionLayer(
301 1U, 1U, conv_filt,
302 get_weights_accessor(data_path, total_path + "pointwise_weights.npy", DataLayout::NCHW),
303 std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
304 conv_pad_stride_info)
305 .set_name(total_path + "pointwise/Conv2D")
306 << BatchNormalizationLayer(
307 get_weights_accessor(data_path, total_path + "pointwise_BatchNorm_moving_mean.npy"),
308 get_weights_accessor(data_path, total_path + "pointwise_BatchNorm_moving_variance.npy"),
309 get_weights_accessor(data_path, total_path + "pointwise_BatchNorm_gamma.npy"),
310 get_weights_accessor(data_path, total_path + "pointwise_BatchNorm_beta.npy"),
311 0.001f)
312 .set_name(total_path + "pointwise/BatchNorm")
313 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::BOUNDED_RELU, 6.f)).set_name(total_path + "pointwise/Relu6");
314
315 return ConcatLayer(std::move(sg));
316 }
317
get_dwsc_node_qasymm(const std::string & data_path,std::string && param_path,const unsigned int conv_filt,PadStrideInfo dwc_pad_stride_info,PadStrideInfo conv_pad_stride_info,QuantizationInfo depth_weights_quant_info,QuantizationInfo point_weights_quant_info)318 ConcatLayer get_dwsc_node_qasymm(const std::string &data_path, std::string &¶m_path,
319 const unsigned int conv_filt,
320 PadStrideInfo dwc_pad_stride_info, PadStrideInfo conv_pad_stride_info,
321 QuantizationInfo depth_weights_quant_info, QuantizationInfo point_weights_quant_info)
322 {
323 std::string total_path = param_path + "_";
324 SubStream sg(graph);
325
326 sg << DepthwiseConvolutionLayer(
327 3U, 3U,
328 get_weights_accessor(data_path, total_path + "depthwise_weights.npy"),
329 get_weights_accessor(data_path, total_path + "depthwise_bias.npy"),
330 dwc_pad_stride_info, 1, std::move(depth_weights_quant_info))
331 .set_name(total_path + "depthwise/depthwise")
332 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::LU_BOUNDED_RELU, 6.f)).set_name(total_path + "depthwise/Relu6")
333 << ConvolutionLayer(
334 1U, 1U, conv_filt,
335 get_weights_accessor(data_path, total_path + "pointwise_weights.npy"),
336 get_weights_accessor(data_path, total_path + "pointwise_bias.npy"),
337 conv_pad_stride_info, 1, std::move(point_weights_quant_info))
338 .set_name(total_path + "pointwise/Conv2D")
339 << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::LU_BOUNDED_RELU, 6.f)).set_name(total_path + "pointwise/Relu6");
340
341 return ConcatLayer(std::move(sg));
342 }
343 };
344
345 /** Main program for MobileNetV1
346 *
347 * Model is based on:
348 * https://arxiv.org/abs/1704.04861
349 * "MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications"
350 * Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, Hartwig Adam
351 *
352 * Provenance: download.tensorflow.org/models/mobilenet_v1_2018_08_02/mobilenet_v1_1.0_224.tgz
353 * download.tensorflow.org/models/mobilenet_v1_2018_08_02/mobilenet_v1_0.75_160.tgz
354 *
355 * @note To list all the possible arguments execute the binary appended with the --help option
356 *
357 * @param[in] argc Number of arguments
358 * @param[in] argv Arguments
359 */
main(int argc,char ** argv)360 int main(int argc, char **argv)
361 {
362 return arm_compute::utils::run_example<GraphMobilenetExample>(argc, argv);
363 }
364