• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1#!/usr/bin/env python3
2# Copyright © 2020 NXP and Contributors. All rights reserved.
3# SPDX-License-Identifier: MIT
4
5import example_utils as eu
6import os
7
8if __name__ == "__main__":
9    args = eu.parse_command_line()
10
11    # names of the files in the archive
12    labels_filename = 'labels_mobilenet_quant_v1_224.txt'
13    model_filename = 'mobilenet_v1_1.0_224_quant.tflite'
14    archive_filename = 'mobilenet_v1_1.0_224_quant_and_labels.zip'
15
16    archive_url = \
17        'https://storage.googleapis.com/download.tensorflow.org/models/tflite/mobilenet_v1_1.0_224_quant_and_labels.zip'
18
19    model_filename, labels_filename = eu.get_model_and_labels(args.model_dir, model_filename, labels_filename,
20                                                              archive_filename, archive_url)
21
22    image_filenames = eu.get_images(args.data_dir)
23
24    # all 3 resources must exist to proceed further
25    assert os.path.exists(labels_filename)
26    assert os.path.exists(model_filename)
27    assert image_filenames
28    for im in image_filenames:
29        assert(os.path.exists(im))
30
31    # Create a network from the model file
32    net_id, graph_id, parser, runtime = eu.create_tflite_network(model_filename)
33
34    # Load input information from the model
35    # tflite has all the need information in the model unlike other formats
36    input_names = parser.GetSubgraphInputTensorNames(graph_id)
37    assert len(input_names) == 1  # there should be 1 input tensor in mobilenet
38
39    input_binding_info = parser.GetNetworkInputBindingInfo(graph_id, input_names[0])
40    input_width = input_binding_info[1].GetShape()[1]
41    input_height = input_binding_info[1].GetShape()[2]
42
43    # Load output information from the model and create output tensors
44    output_names = parser.GetSubgraphOutputTensorNames(graph_id)
45    assert len(output_names) == 1  # and only one output tensor
46    output_binding_info = parser.GetNetworkOutputBindingInfo(graph_id, output_names[0])
47
48    # Load labels file
49    labels = eu.load_labels(labels_filename)
50
51    # Load images and resize to expected size
52    images = eu.load_images(image_filenames, input_width, input_height)
53
54    eu.run_inference(runtime, net_id, images, labels, input_binding_info, output_binding_info)
55