Bilityuk's profile - activity

Exception 0x7B17A5A9 (opencv_core2410.dll) in sdkDemo.exe: 0xC0000005: acces violation writing the address 0x00000001. H

Diferent objects detection in one picture Hi, i am trying to find the right way of how to detect different objects in on

Diferent objects detection in one picture Hi, i am trying to find the right way of how to detect different objects in on

Can anybody help me please?

I tried to use openMP to reduce work time of function, but result became worse. it takes 70-75 msec if i use openMP for predict:

#pragma omp parallel
{
    model->predict(preprocessedFace, label, confidence); 
}

No i mean one person. Firstly i run the code with 10 faces for one person and it took 3 msec, then i run 200 faces for the same person and it took 40-45 msec.

For 10 faces it takes 3 msec For 200 faces it takes 40-45 msec

I am not sure.. Ok i understand the task.. I will measure the time of predict function with 10 faces in training set and with 200 faces. And let you know the results..

Hello, can anybody share a solution how to reduce operation time for LBPHFaceRecognizer model->predict() method? When i load more than 100 faces to the trainset and after training start the prediction loop, it works very slow. How to increase the perfomance?

Thank you it works for me

Good morning! Can anybody help me to convert between YUYV422 and BGR pixel formats?

1) This way i load the pixel data: cv::Mat cameraFrame = cv::Mat(h, w,CV_8UC3,(char*)YUYV422_data); 
2) i need to convert cv::Mat YUYV422  to cv::Mat BGR 
3) after processing i need to convert cv::Mat BGR to (char*)YUYV422_data.

Thank you!

Can anybody share a working pixel convertion function. I cant find the right solution fo my problem..

The broblem is clear for now. How to convert BGR24 to YUYV422 that i can draw it with DrawVideoImage function? Wich CV_8U option to use (CV_8UC1,CV_8UC2,CV_8UC3,CV_8UC4) and wich option to use for cv::CvtColor(dst,src, ????).

The second part of the problem is in:

Mat InImg(h, w,CV_8UC1);  
memcpy(InImg.data,Frame->data, sizeof(char)*w*h);           
InImg.copyTo(displayedFrame);

Becase i think it is not properly copy the char* data from Frame to Mat, maybe it requires some decoding i have no idea. If i put the displayedFrame to imshow() just after i memcpy() it to Mat it will not be drawn properly.

I feel that the problem somewhere in (char*)displayedFrame.data. I think that the renderer doesnt recognize it as native Frame->data format that is of type char data[0];

I don't think copying to a Mat is your problem, but getting the right data. When I look at the Frame struct, I see that there's more than just the data container. So I think GetVideoData(hwnd, (char)&Frame, &dataLen, videofmt,&mediaSpeed); could be the problem. At first I would check (char)&Frame and try something like &Frame->data

I am sure that

GetVideoData(hwnd, (char**)&Frame, &dataLen, videofmt,&mediaSpeed);

is not a problem, because it works perfect and if i pass the native Frame->data to the renderer it draws the video even with 30 fps without a problems..

m_d3d.DrawVideoIamge(Frame->data, 1, 0);

But when i am trying to pass the processed Mat to renderer there is a problem:

m_d3d.DrawVideoIamge((char*)displayedFrame.data, 1, 0);

Here is the rest of the enum:

DECTYPE_YUV420,
DECTYPE_YUYV422,
DECTYPE_UYVY422,
DECTYPE_H264,
DECTYPE_MJPEG,  
DECTYPE_MJPEG_BASE64,
DECTYPE_H264_BASE64,        
DECTYPE_AUDIORAW,
DECTYPE_G726,
DECTYPE_G711U,
DECTYPE_PCM,
DECTYPE_ADPCM,
DECTYPE_G711A,
DECTYPE_AAC,
}DECFMT;

berak, can you explaine please wich way to dig? DECFMT has the following types:

typedef enum  
{
DECTYPE_VIDEORAW,
DECTYPE_ARGB32,   //packed ARGB 8:8:8:8, 32bpp, ARGBARGB...
DECTYPE_RGBA32,    //packed RGBA 8:8:8:8, 32bpp, RGBARGBA...
DECTYPE_ABGR32,    //packed ABGR 8:8:8:8, 32bpp, ABGRABGR...
DECTYPE_BGRA32,    //packed BGRA 8:8:8:8, 32bpp, BGRABGRA...
DECTYPE_RGB24,    //packed RGB 8:8:8, 24bpp, RGBRGB...
DECTYPE_BGR24,     //packed RGB 8:8:8, 24bpp, BGRBGR...
DECTYPE_RGB565BE,  //packed RGB 5:6:5, 16bpp, (msb)   5R 6G 5B(lsb), big-endian
DECTYPE_RGB565LE,  //packed RGB 5:6:5, 16bpp, (msb)   5R 6G 5B(lsb), little-endian
DECTYPE_BGR565BE,  //packed BGR 5:6:5, 16bpp, (msb)   5B 6G 5R(lsb), big-endian
DECTYPE_BGR565LE,  //packed BGR 5:6:5, 16bpp

could be the problem. At first I would check (char**)&Frame and try something like &Frame->data

&Frame->data doesnt make any sence in compare with Frame->data The same result.

Function:

if (GetVideoData(hFos, (char**)&Frame, &dataLen, videofmt,&mediaSpeed) )
{
SetMediaSpeed(mediaSpeed);
if (dataLen)
{
w = pFrame->media.video.picWidth;
h = pFrame->media.video.picHeight;
m_ddraw.CreatEmptySurface(Frame->media.video.picWidth, Frame->media.video.picHeight);
m_ddraw.CreatOverlaySurface(Frame->media.video.picWidth, Frame->media.video.picHeight);
HWND hWnd = GetDlgItem(IDC_STATIC_SCREEN)->GetSafeHwnd();
m_d3d.InitDDraw(hWnd, w, h);
}
Mat InImg(h, w,CV_8UC3);  
memcpy(InImg.data,pFrame->data, sizeof(char)*w*h);          
InImg.copyTo(displayedFrame);

//Face Recogition processing//<<-- process the displayedFrame Mat

//display image//
m_d3d.DrawVideoIamge((char*)displayedFrame.data, 1, 0);
}

Here is global variables:

HANDLE hFos = m_h;
DEC_DATA* Frame = NULL;
int dataLen = 0;
DECFMT videofmt = DECTYPE_YUYV422;
DECFMT audiofmt = DECTYPE_PCM;
int w=0, h=0;
int mediaSpeed = 0;

I want to post the full function code. But there is not enough space for comments. I will post it in slices

Just tried:

Mat cameraFrame;
Mat InImg(h, w,CV_8UC3);  
memcpy(InImg.data,Frame->data, 3*w*h);       
InImg.copyTo(cameraFrame);

It throws an exception on the memcpy line of code:

CV_8UC3 doesn't work for this code(throws c000005):

cv::Mat cameraFrame = cv::Mat(h, w, CV_8UC3, Frame->data);

If use CV_8UC3 with this code:

Mat InImg(h, w,CV_8UC3);  
memcpy(InImg.data,pFrame->data, sizeof(char)*w*h);  
InImg.copyTo(cameraFrame);

Results the almost the same as for CV_8UC1. But now it lookes like triple image.

Here is CV_8UC3 with imshow():

Here is CV_8UC3 with DrawVideoIamge():

If i use imshow() instead of DrawVideoIamge(). I've got the gray image like on the previos screen i posted.

Also i tried to use another way of convertion, and it behaves different. Here i convert the Frame->data to cv:Mat:

Mat cameraFrame;
Mat InImg(h, w,CV_8UC1);  
memcpy(InImg.data,Frame->data, sizeof(char)*w*h);       
InImg.copyTo(cameraFrame);

make some face recognition processing with cameraFrame and finally trying to draw the image this way:

DrawVideoIamge((char*)cameraFrame.data, 1, 0);

And now i got the color image, but it seems like it mirrored or i dont know how to explaine. Here is a link i upload the screen shot.

</a">link text