azONNXWrapper.cpp
1.1 KB
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
#include "azONNXWrapper.h"
int AZONNXWrapper::init(const std::string& model_path)
{
if (!model_path.empty()) {
//加载并初始化网络
net_ = cv::dnn::readNetFromONNX(model_path);
net_.setPreferableBackend(cv::dnn::Backend::DNN_BACKEND_OPENCV);
net_.setPreferableTarget(cv::dnn::Target::DNN_TARGET_CPU);
return 0;
}
return -1;
}
float* AZONNXWrapper::forward(cv::Mat img, cv::Scalar mean, cv::Scalar std)
{
CV_Assert(!img.empty());
//图像预处理,仅支持三通道
cv::Mat input;
int incn = img.channels();
if (incn < 3) {
cv::cvtColor(img, input, cv::COLOR_GRAY2RGB);
}
else if (incn > 3) {
cv::cvtColor(img, input, cv::COLOR_BGR2RGB);
}
else {
input = img.clone();
}
//转float类型
input.convertTo(input, CV_32F, 1 / 255.);
//预处理
cv::subtract(input, mean, input);
cv::divide(input, std, input);
//设定输出
cv::Mat blob = cv::dnn::blobFromImage(input);
net_.setInput(blob);
//推理
cv::Mat predicts = net_.forward();
//
for (int i = 0; i < 512; i++) predictions[i] = predicts.ptr<float>(0)[i];
return predictions;
}
AZONNXWrapper::AZONNXWrapper() {
}
AZONNXWrapper::~AZONNXWrapper()
{
delete[] predictions;
predictions = NULL;
}