1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
|
/*
* Copyright (c) 2018 Samsung Electronics Co., Ltd. All Rights Reserved
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include "ImageClassifier.h"
#include <fstream>
#include <queue>
#include <algorithm>
ImageClassifier::ImageClassifier(const std::string &model_file, const std::string &label_file,
const int input_size, const int image_mean, const int image_std,
const std::string &input_name, const std::string &output_name,
const bool use_nnapi)
: _inference(new InferenceInterface(model_file, use_nnapi)), _input_size(input_size),
_image_mean(image_mean), _image_std(image_std), _input_name(input_name),
_output_name(output_name)
{
// Load label
std::ifstream label_stream(label_file.c_str());
assert(label_stream);
std::string line;
while (std::getline(label_stream, line))
{
_labels.push_back(line);
}
_num_classes = _inference->getTensorSize(_output_name);
std::cout << "Output tensor size is " << _num_classes << ", label size is " << _labels.size()
<< std::endl;
// Pre-allocate buffers
_fdata.reserve(_input_size * _input_size * 3);
_outputs.reserve(_num_classes);
}
std::vector<Recognition> ImageClassifier::recognizeImage(const cv::Mat &image)
{
// Resize image
cv::Mat cropped;
cv::resize(image, cropped, cv::Size(_input_size, _input_size), 0, 0, cv::INTER_AREA);
// Preprocess the image data from 0~255 int to normalized float based
// on the provided parameters
_fdata.clear();
for (int y = 0; y < cropped.rows; ++y)
{
for (int x = 0; x < cropped.cols; ++x)
{
cv::Vec3b color = cropped.at<cv::Vec3b>(y, x);
color[0] = color[0] - (float)_image_mean / _image_std;
color[1] = color[1] - (float)_image_mean / _image_std;
color[2] = color[2] - (float)_image_mean / _image_std;
_fdata.push_back(color[0]);
_fdata.push_back(color[1]);
_fdata.push_back(color[2]);
cropped.at<cv::Vec3b>(y, x) = color;
}
}
// Copy the input data into model
_inference->feed(_input_name, _fdata, 1, _input_size, _input_size, 3);
// Run the inference call
_inference->run(_output_name);
// Copy the output tensor back into the output array
_inference->fetch(_output_name, _outputs);
// Find the best classifications
auto compare = [](const Recognition &lhs, const Recognition &rhs) {
return lhs.confidence < rhs.confidence;
};
std::priority_queue<Recognition, std::vector<Recognition>, decltype(compare)> pq(compare);
for (int i = 0; i < _num_classes; ++i)
{
if (_outputs[i] > _threshold)
{
pq.push(Recognition(_outputs[i], _labels[i]));
}
}
std::vector<Recognition> results;
int min = std::min(pq.size(), _max_results);
for (int i = 0; i < min; ++i)
{
results.push_back(pq.top());
pq.pop();
}
return results;
}
|
