void VideoGeneration::on_generateVideoPushButton_clicked()
{
//string camPath = "C:\\Users\\dehandecroos\\Desktop\\Videos\\PRG28.avi";
QString profileId = ui.profileName_lineEdit->text();
string cameraIds[] = {
"camera_node_6_log",
"camera_node_1_log",
"camera_node_28_log",
"camera_node_23_log"
};
int cameraIdsSize = sizeof(cameraIds) / sizeof(*cameraIds);
string finalJoinQuery = "";
int i = 1;
for (string cameraId : cameraIds)
{
finalJoinQuery += "select * from " + cameraId + " where profile_id='" + profileId.toStdString() + "'";
if (i++ != cameraIdsSize) {
finalJoinQuery += "union ";
}
}
finalJoinQuery += "order by TimeStamp";
struct CameraTimeStamp{
string cameraId;
double timestamp;
};
stmt->execute("USE camera");
ResultSet *timeStampsForProfile = stmt->executeQuery(finalJoinQuery);
vector<CameraTimeStamp> timeStamps;
while (timeStampsForProfile->next())
{
CameraTimeStamp timeStamp;
timeStamp.cameraId = timeStampsForProfile->getString("Video_ID");
timeStamp.timestamp = timeStampsForProfile->getDouble("TimeStamp");
timeStamps.push_back(timeStamp);
}
vector<Mat> video;
for (CameraTimeStamp ts : timeStamps)
{
string camPath = "C:\\AdaptiveCameraNetworkPack\\Videos\\";
string camId = ts.cameraId;
camPath += "PRG" + camId + ".avi";
VideoCapture cam;
cam.open(camPath);
int frameRate = cam.get(CV_CAP_PROP_FPS);
int minutes = (int)ts.timestamp;
int seconds = (int)((ts.timestamp-(double)minutes)*100.0);
int milis = ((ts.timestamp - (double)minutes)*100.0-seconds)*1000;
int milliseconds = (minutes * 60 + seconds) * 1000 + milis;
qDebug() << "Extracted Frames for time " + QString::number(ts.timestamp) + ", in camera " + QString::fromStdString(camId);
cam.set(CV_CAP_PROP_POS_MSEC, milliseconds);
for (int frameCount = 0; frameCount < frameRate; frameCount++)
{
Mat frame;
cam >> frame;
if (frame.empty())
{
break;
}
int fontFace = FONT_HERSHEY_SIMPLEX;
double fontScale = 1;
int thickness = 3;
cv::Point textOrg1(10, 50);
putText(frame, "CAM:" + ts.cameraId, textOrg1, fontFace, fontScale, Scalar::all(0),2);
cv::Point textOrg2(500, 50);
video.push_back(frame);
}
//VideoCapture
}
if (video.size() == 0){
QImage finalImage(ui.lblOutput->width(), ui.lblOutput->width(), QImage::Format_RGB888);
QPainter qp(&finalImage);
qp.setBrush(Qt::black);
qp.setPen(Qt::red);
qp.setFont(QFont("Times", 12, QFont::Bold));
qp.drawText(finalImage.rect(), Qt::AlignCenter, "NO VIDEO FOR "+ profileId);
ui.lblOutput->setPixmap(QPixmap::fromImage(finalImage));
}
else
{
for (Mat frameZ : video)
{
Mat frameForQimage;
cvtColor(frameZ, frameForQimage, CV_BGR2RGB);
QImage outImage((uchar*)frameForQimage.data, frameForQimage.cols, frameForQimage.rows, frameZ.step, QImage::Format_RGB888);
ui.lblOutput->setPixmap(QPixmap::fromImage(outImage));
imshow("Output", frameZ);
cvWaitKey(1);
}
}
}
void PM_dem::detect( Mat &img, float score_thresh, bool show_hints, bool show_img, string save_img )
{
if( score_thresh==DEFAULT_THRESH )
model.threshing = model.thresh;
else
model.threshing = score_thresh;
hints = show_hints;
// 1. Feature pyramid <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
prag_start = yuGetCurrentTime('M');
if( hints ){
printf("Calculating feature pyramid ...\n");
start_clock = prag_start;
}
featpyramid2( img, model, pyra );
if( hints ){
end_clock = yuGetCurrentTime('M');
printf("Time for _featpyramid is %gs\n",(end_clock-start_clock)/1000.f);
}
// 2. Compute PCA projection of the feature pyramid <<<<<<<<<<<<<<
if( hints ){
printf("Compute PCA projection of the feature pyramid ...\n");
start_clock = end_clock;
}
/*Mat Ctest = model.C2(Rect(0,155,32,1));
cout<<Ctest;*/
//project_pyramid( model, pyra );
if( hints ){
end_clock = yuGetCurrentTime('M');
printf("Time for _project_pyramid() is %gs\n",(end_clock-start_clock)/1000.f);
}
if (hints)
{
end_clock = start_clock;
printf("QT\n");
}
qtpyra(model,pyra);
if (hints)
{
end_clock = yuGetCurrentTime('M');
printf("%gs\n",(end_clock-start_clock)/1000.f);
}
if( pyra.num_levels!=pyra.feat.size() ){
printf("pyra.num_levels!=pyra.feat.size()\n");
throw runtime_error("");
}
// 3. Precompute location/scale scores <<<<<<<<<<<<<<<<<<<<<<<
Mat loc_f = loc_feat( model, pyra.num_levels );
pyra.loc_scores.resize( model.numcomponents );
for( int c=0; c<model.numcomponents; c++ ){
Mat loc_w( 1, model.loc[c].w.size(), CV_32FC1, &(model.loc[c].w[0]) ); // loc_w = model.loc[c].w
pyra.loc_scores[c] = loc_w * loc_f;
}
// 4. Gather PCA root filters for convolution <<<<<<<<<<<<<<<<<<<
if( hints ){
printf("Gathering PCA root filters for convolution ...\n");
start_clock = end_clock;
}
if( rootscores[0].size()!=pyra.num_levels ){
vector<Mat> tmp_rootscores(pyra.num_levels);
rootscores.assign(model.numcomponents,tmp_rootscores);
}
// ofstream f1("pj.txt");
int numrootlocs = 0;
int s = 0; // will hold the amount of temp storage needed by cascade()
for( int i=0; i<pyra.num_levels; i++ ){
s += pyra.feat[i].rows * pyra.feat[i].cols;
if( i<model.interval )
continue;
static vector<Mat> scores;
//fconv( pyra.projfeat[i], rootfilters, 0, numrootfilters, scores );
fconv_root_qt(model, pyra.qtfeat[i], scores);
for( int c=0; c<model.numcomponents; c++ ){
int u = model.components[c].rootindex;
int v = model.components[c].offsetindex;
float tmp = model.offsets[v].w + pyra.loc_scores[c].at<float>(i);
rootscores[c][i] = scores[u] + Scalar(tmp);
numrootlocs += scores[u].total();
}
}
cout<<numrootlocs<<endl;
s = s * model.partfilters.size();
if( hints ){
end_clock = yuGetCurrentTime('M');
printf("Time for gathering PCA root filters is %gs\n",(end_clock-start_clock)/1000.f);
}
// 5. Cascade detection in action <<<<<<<<<<<<<<<<<<<<<<<
if( hints ){
printf("Cascade detection in action ...\n");
start_clock = end_clock;
}
//Mat coords = cascade(model, pyra, rootscores, numrootlocs, s);
Mat coords = cascade_qt(model, pyra, rootscores, numrootlocs, s);
//cout<<coords;
//cout<<"??"<<endl;
if( hints ){
end_clock = yuGetCurrentTime('M');
printf("Time for _cascade() is %gs\n",(end_clock-start_clock)/1000.f);
if( coords.empty() ){
printf("No Detection!\n");
return;
}
}
// 6. Detection results <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
Mat boxes = getboxes( model, img_color, coords );
Mat x1 = boxes.col(0);
Mat y1 = boxes.col(1);
Mat x2 = boxes.col(2);
Mat y2 = boxes.col(3);
Mat Score = boxes.col( boxes.cols-1 );
detections.resize( x1.rows );
for( int i=0; i<x1.rows; i++ ){
detections[i][0] = x1.at<float>(i);
detections[i][1] = y1.at<float>(i);
detections[i][2] = x2.at<float>(i);
detections[i][3] = y2.at<float>(i);
detections[i][4] = Score.at<float>(i);
}
if( hints ){
prag_end = yuGetCurrentTime('M');
printf("Total detection time is : %gs\n",(prag_end-prag_start)/1000.f);
}
// 6. Draw and show <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
if( show_img || !save_img.empty() ){
//showboxes( img_color, boxes );
//
const int fontFace = CV_FONT_HERSHEY_PLAIN;
const double fontScale = 1;
const Scalar drawColor = CV_RGB(255,0,0);
const Scalar fontColor = CV_RGB(30,250,150);
//
for( int i=0; i!=detections.size(); i++ ){
float x1 = detections[i][0], y1 = detections[i][1], x2 = detections[i][2], y2 = detections[i][3];
float _score = detections[i][4];
//
Point2f UL( x1, y1 );
Point2f BR( x2, y2 );
rectangle( img_color, UL, BR, drawColor, 2 );
printf("----------------------------\n");
printf("%g %g %g %g %g\n", x1, y1, x2, y2, _score );
//
x1 = int(x1*10+0.5) / 10.f; // ½ö±£Áô1λСÊý
y1 = int(y1*10+0.5) / 10.f;
x2 = int(x2*10+0.5) / 10.f;
y2 = int(y2*10+0.5) / 10.f;
_score = int(_score*100+0.5) / 100.f;
//
char buf[50] = { 0 };
sprintf_s( buf, 50, "%d", i );
string text = buf;
int baseline = 0;
Size textSize = getTextSize( text, fontFace, fontScale, 1, &baseline );
Point2f textOrg2( x1, y1+textSize.height+2 );
putText( img_color, text, textOrg2, fontFace, fontScale, fontColor );
//
sprintf_s( buf, 50, "%d %g %g %g %g %g", i, x1, y1, x2, y2, _score );
text = buf;
textSize = getTextSize( text, fontFace, fontScale, 1, &baseline );
Point2f textOrg(5,(i+1)*(textSize.height+3));
putText( img_color, text, textOrg, fontFace, fontScale, fontColor );
}
{
char buf[30] = { 0 };
sprintf_s( buf, 30, "time : %gs", (prag_end-prag_start)/1000.f );
string time_text = buf;
int baseline = 0;
Size textSize = getTextSize( time_text, fontFace, fontScale, 1, &baseline );
Point2f time_orig(5,(detections.size()+1)*(textSize.height+3));
putText( img_color, time_text, time_orig, fontFace, fontScale, fontColor );
}
if( show_img )
imshow( "OK", img_color );
if( !save_img.empty() )
imwrite( save_img, img_color );
}
}
void Objectness::PCANet_pred(Mat img, vector<Mat> &testImg, vector<Rect> &bboxes, PCA_Train_Result* result, PCANet pcaNet){
// cout << "\n ====== PCANet Testing ======= \n" << endl;
int testSIze = testImg.size();
// cout << testSIze << endl;
Hashing_Result* hashing_r;
PCA_Out_Result *out;
PCA_Out_Result *res;
int coreNum = omp_get_num_procs();
#pragma omp parallel for default(none) num_threads(coreNum) private(out, res,hashing_r) shared(mySVM,libModel, pcaNet, testSIze, testImg, result, bboxes,img,color)
for (int i = 0; i < testSIze; i++){
out = new PCA_Out_Result;
out->OutImgIdx.push_back(0);
out->OutImg.push_back(testImg[i]);
PCA_output(res,out->OutImg, out->OutImgIdx, pcaNet.PatchSize,
pcaNet.NumFilters[0], result->Filters[0], 2);
for (int j = 1; j < pcaNet.NumFilters[1]; j++)
res->OutImgIdx.push_back(j);
delete out;
out = new PCA_Out_Result;
PCA_output(out, res->OutImg, res->OutImgIdx, pcaNet.PatchSize,
pcaNet.NumFilters[1], result->Filters[1], 2);
HashingHist(hashing_r,&pcaNet, out->OutImgIdx, out->OutImg);
hashing_r->Features.convertTo(hashing_r->Features, CV_32F);
delete res;
delete out;
// OpenCV SVM prediction
float pred = mySVM.predict(hashing_r->Features/*, true*/);
/*========== liblinear prediction =================*/
//struct feature_node *svmVec;
//svmVec = (struct feature_node *) malloc((hashing_r->Features.cols + 1)*sizeof(struct feature_node));
//double *predictions = new double[hashing_r->Features.rows];
//float *dataPtr = hashing_r->Features.ptr<float>(); // Get data from OpenCV Mat
//double prob_est[2]; // Probability estimation
//int r, c;
//for (r = 0; r < hashing_r->Features.rows; r++)
//{
// for (c = 0; c < hashing_r->Features.cols; c++)
// {
// svmVec[c].index = c + 1; // Index starts from 1; Pre-computed kernel starts from 0
// svmVec[c].value = dataPtr[r*hashing_r->Features.cols + c];
// }
// svmVec[c].index = -1; // End of line
//}
//double pred = predict(libModel, svmVec);
//free(svmVec);
//printf("pred: %f\n", pred);
delete hashing_r;
#pragma omp critical
if (pred < 3){
//cout << imgnames[i] << endl;
Rect box = bboxes[i];
//printf("predict: %f\n", pred);
/*========== save to patch =============*/
//cv::Mat croppedImage;
//img(box).copyTo(croppedImage);
////imshow("asd", croppedImage);
//char* patch_name = new char[30];
//sprintf(patch_name, "%s%d%s", "patch/", k + 1, ".jpg");
//imwrite(patch_name, croppedImage);
//k++;
rectangle(img, cvPoint(box.x, box.y), cvPoint(box.x + box.width, box.y + box.height), color, 1);
// ========= add some text ============
char buf[50] = { 0 };
if (pred == 1)
sprintf_s(buf, 50, "right");
else if (pred == 2)
sprintf_s(buf, 50, "forward");
else
sprintf_s(buf, 50, "left");
string text = buf;
int baseline = 0;
Size textSize = getTextSize(text, fontFace, fontScale, 1, &baseline);
Point2f textOrg2(box.x, box.y + textSize.height + 2);
putText(img, text, textOrg2, fontFace, fontScale, fontColor);
}
}
}
