no message
parent
f748b3ded8
commit
ed2d2e9e09
|
|
@ -76,20 +76,21 @@ void CameraVideoSink::OnFrame(const webrtc::VideoFrame& frame) {
|
|||
std::chrono::system_clock::now().time_since_epoch()).count();
|
||||
static size_t cnt = 0;
|
||||
|
||||
qDebug()<<int(frame.video_frame_buffer()->type());
|
||||
// qDebug()<<int(frame.video_frame_buffer()->type());
|
||||
rtc::scoped_refptr<webrtc::I420BufferInterface> frameBuffer =
|
||||
frame.video_frame_buffer()->ToI420();
|
||||
cnt++;
|
||||
int width = this->m_capability.width;
|
||||
int height = this->m_capability.height;
|
||||
/*
|
||||
qDebug()<<this->Capability().height<<this->Capability().width
|
||||
<<int(this->Capability().videoType)
|
||||
<<int(webrtc::VideoType::kI420)
|
||||
<<this->Capability().interlaced<<width * height*4;
|
||||
<<this->Capability().interlaced<<width * height*4;*/
|
||||
|
||||
uint8_t *data =
|
||||
new uint8_t[width * height*4];
|
||||
qDebug()<<width*height<<int(frameBuffer->GetI420()->type());
|
||||
// qDebug()<<width*height<<int(frameBuffer->GetI420()->type());
|
||||
memcpy(data,frameBuffer->GetI420()->DataY(),width*height);
|
||||
memcpy(data + width*height ,frameBuffer->GetI420()->DataU(),
|
||||
width*height/4);
|
||||
|
|
|
|||
|
|
@ -13,29 +13,44 @@ String objNames[] = { "background",
|
|||
"motorbike", "person", "pottedplant",
|
||||
"sheep", "sofa", "train", "tvmonitor" };
|
||||
|
||||
|
||||
cv::Mat * ssd_detect(cv::Mat *inframe) {
|
||||
if (inframe->empty()) {
|
||||
printf("could not load image...\n");
|
||||
return inframe;
|
||||
}
|
||||
namedWindow("input image", WINDOW_AUTOSIZE);
|
||||
imshow("input image", *inframe);
|
||||
|
||||
Net net = readNetFromCaffe(model_text_file, modelFile);
|
||||
|
||||
// image：这个就是我们将要输入神经网络进行处理或者分类的图片。
|
||||
// mean：需要将图片整体减去的平均值，如果我们需要对RGB图片的三个通道分别减去不同的值，那么可以使用3组平均值，如果只使用一组，那么就默认对三个通道减去一样的值。减去平均值（mean）：为了消除同一场景下不同光照的图片，对我们最终的分类或者神经网络的影响，我们常常对图片的R、G、B通道的像素求一个平均值，然后将每个像素值减去我们的平均值，这样就可以得到像素之间的相对值，就可以排除光照的影响。
|
||||
// scalefactor：当我们将图片减去平均值之后，还可以对剩下的像素值进行一定的尺度缩放，它的默认值是1，如果希望减去平均像素之后的值，全部缩小一半，那么可以将scalefactor设为1/2。
|
||||
// size：这个参数是我们神经网络在训练的时候要求输入的图片尺寸。
|
||||
// swapRB：OpenCV中认为我们的图片通道顺序是BGR，但是我平均值假设的顺序是RGB，所以如果需要交换R和G，那么就要使swapRB=true
|
||||
Mat blobImage = blobFromImage(*inframe, 0.007843,
|
||||
cv::Mat resize;
|
||||
cv::resize(*inframe, resize,
|
||||
Size(480,560),
|
||||
0, 0, INTER_LINEAR);// X Y各缩小一半
|
||||
qDebug()<<"frame.type()"<<resize.type();
|
||||
cv::Mat image2(480,640,CV_8UC3);
|
||||
qDebug()<<"image2 ()"<<image2.type()<<CV_8UC3;
|
||||
cv::cvtColor(resize,image2,COLOR_RGBA2RGB);
|
||||
imshow("input image", image2);
|
||||
qDebug()<<"image2 frame.type()"<<image2.type()<<CV_8UC3;
|
||||
Mat blobImage1 = blobFromImage(image2, 0.007843,
|
||||
Size(300, 300),
|
||||
Scalar(127.5, 127.5, 127.5), true, false);
|
||||
qDebug()<<"blobImage width : " << blobImage.cols
|
||||
<<"blobImage.cols: " << blobImage.rows;
|
||||
|
||||
imshow("ssd-demo", blobImage);
|
||||
// Net net = readNetFromCaffe(model_text_file, modelFile);
|
||||
|
||||
//// image：这个就是我们将要输入神经网络进行处理或者分类的图片。
|
||||
//// mean：需要将图片整体减去的平均值，如果我们需要对RGB图片的三个通道分别减去不同的值，那么可以使用3组平均值，
|
||||
//// 如果只使用一组，那么就默认对三个通道减去一样的值。减去平均值（mean）：为了消除同一场景下不同光照的图片，
|
||||
//// 对我们最终的分类或者神经网络的影响，我们常常对图片的R、G、B通道的像素求一个平均值，
|
||||
//// 然后将每个像素值减去我们的平均值，这样就可以得到像素之间的相对值，就可以排除光照的影响。
|
||||
//// scalefactor：当我们将图片减去平均值之后，还可以对剩下的像素值进行一定的尺度缩放，
|
||||
//// 它的默认值是1，如果希望减去平均像素之后的值，全部缩小一半，那么可以将scalefactor设为1/2。
|
||||
//// size：这个参数是我们神经网络在训练的时候要求输入的图片尺寸。
|
||||
//// swapRB：OpenCV中认为我们的图片通道顺序是BGR，但是我平均值假设的顺序是RGB，
|
||||
/// 所以如果需要交换R和G，那么就要使swapRB=true
|
||||
// cv::Mat blobImage = blobFromImage(resize,
|
||||
// 0.007843,
|
||||
// Size(300, 300),
|
||||
// Scalar(127.5, 127.5, 127.5), true, false);
|
||||
// qDebug()<<"blobImage width : " << blobImage.cols
|
||||
// <<"blobImage.cols: " << blobImage.rows;
|
||||
|
||||
// net.setInput(blobImage, "data");
|
||||
// Mat detection = net.forward("detection_out");
|
||||
|
|
@ -45,20 +60,58 @@ cv::Mat * ssd_detect(cv::Mat *inframe) {
|
|||
// float confidence = detectionMat.at<float>(i, 2);
|
||||
// if (confidence > confidence_threshold) {
|
||||
// size_t objIndex = (size_t)(detectionMat.at<float>(i, 1));
|
||||
// float tl_x = detectionMat.at<float>(i, 3) * frame->cols;
|
||||
// float tl_y = detectionMat.at<float>(i, 4) * frame->rows;
|
||||
// float br_x = detectionMat.at<float>(i, 5) * frame->cols;
|
||||
// float br_y = detectionMat.at<float>(i, 6) * frame->rows;
|
||||
// float tl_x = detectionMat.at<float>(i, 3) * resize.cols;
|
||||
// float tl_y = detectionMat.at<float>(i, 4) * resize.rows;
|
||||
// float br_x = detectionMat.at<float>(i, 5) * resize.cols;
|
||||
// float br_y = detectionMat.at<float>(i, 6) * resize.rows;
|
||||
|
||||
// Rect object_box((int)tl_x, (int)tl_y, (int)(br_x - tl_x), (int)(br_y - tl_y));
|
||||
// rectangle(*frame, object_box, Scalar(0, 0, 255), 2, 8, 0);
|
||||
// putText(*frame, format("%s", objNames[objIndex].c_str()), Point(tl_x, tl_y),
|
||||
// rectangle(resize, object_box, Scalar(0, 0, 255), 2, 8, 0);
|
||||
// putText(resize, format("%s", objNames[objIndex].c_str()), Point(tl_x, tl_y),
|
||||
// FONT_HERSHEY_SIMPLEX, 1.0, Scalar(255, 0, 0), 2);
|
||||
// }
|
||||
// }
|
||||
// imshow("ssd-demo", resize);
|
||||
|
||||
// waitKey(0);
|
||||
return inframe;
|
||||
////// waitKey(0);
|
||||
// return inframe;
|
||||
Mat frame = imread("D:/project/opencv_tutorial_data/images/gaoyy.png");
|
||||
|
||||
if (frame.empty()) {
|
||||
printf("could not load image...\n");
|
||||
return nullptr;
|
||||
}
|
||||
qDebug()<<"frame.type()"<<frame.type()<<frame.cols<<frame.rows<<frame.depth()
|
||||
<<frame.elemSize();
|
||||
namedWindow("input image", WINDOW_AUTOSIZE);
|
||||
|
||||
Net net = readNetFromCaffe(model_text_file, modelFile);
|
||||
Mat blobImage = blobFromImage(frame, 0.007843,
|
||||
Size(300, 300),
|
||||
Scalar(127.5, 127.5, 127.5), true, false);
|
||||
qDebug()<<"blobImage width "<< blobImage.cols<<blobImage.rows;
|
||||
|
||||
net.setInput(blobImage1, "data");
|
||||
Mat detection = net.forward("detection_out");
|
||||
Mat detectionMat(detection.size[2], detection.size[3], CV_32F, detection.ptr<float>());
|
||||
float confidence_threshold = 0.2;
|
||||
for (int i = 0; i < detectionMat.rows; i++) {
|
||||
float confidence = detectionMat.at<float>(i, 2);
|
||||
if (confidence > confidence_threshold) {
|
||||
size_t objIndex = (size_t)(detectionMat.at<float>(i, 1));
|
||||
float tl_x = detectionMat.at<float>(i, 3) * image2.cols;
|
||||
float tl_y = detectionMat.at<float>(i, 4) * image2.rows;
|
||||
float br_x = detectionMat.at<float>(i, 5) * image2.cols;
|
||||
float br_y = detectionMat.at<float>(i, 6) * image2.rows;
|
||||
|
||||
Rect object_box((int)tl_x, (int)tl_y, (int)(br_x - tl_x), (int)(br_y - tl_y));
|
||||
rectangle(image2, object_box, Scalar(0, 0, 255), 2, 8, 0);
|
||||
putText(image2, format("%s", objNames[objIndex].c_str()), Point(tl_x, tl_y), FONT_HERSHEY_SIMPLEX, 1.0, Scalar(255, 0, 0), 2);
|
||||
}
|
||||
}
|
||||
imshow("ssd-demo", image2);
|
||||
waitKey(0);
|
||||
return nullptr;
|
||||
}
|
||||
|
||||
|
||||
|
|
|
|||
|
|
@ -22,7 +22,7 @@
|
|||
<widget class="QWidget" name="centralwidget">
|
||||
<layout class="QVBoxLayout" name="verticalLayout_2" stretch="1,9">
|
||||
<item>
|
||||
<layout class="QHBoxLayout" name="horizontalLayout" stretch="1,1,1,1,1,0,0,3">
|
||||
<layout class="QHBoxLayout" name="horizontalLayout" stretch="1,1,1,1,1,0,0,0,3">
|
||||
<item>
|
||||
<widget class="QLabel" name="label">
|
||||
<property name="text">
|
||||
|
|
@ -70,6 +70,13 @@
|
|||
</property>
|
||||
</widget>
|
||||
</item>
|
||||
<item>
|
||||
<widget class="QPushButton" name="pushButton_4">
|
||||
<property name="text">
|
||||
<string>动态检测</string>
|
||||
</property>
|
||||
</widget>
|
||||
</item>
|
||||
<item>
|
||||
<spacer name="horizontalSpacer">
|
||||
<property name="orientation">
|
||||
|
|
|
|||
Loading…
Reference in New Issue