Tariq's profile - activity

Is this issues fixed. Which branch contains this fix?

I need to do face model training from the webcam video streams.I am updating the model with new face info. Not sure how this model update strategy would behave for 1000 or more faces?

I am doing face detection on GPU using OpenCV GPU implementation. My code is crashing after 20-50 seconds of execution. Following is the error log:

 OpenCV Error: Gpu API call (NCV Assertion Failed: cudaError_t=4, file=C:/opencv-2.4.6/modules/gpu/src/nvidia/NCVHaarObjectDetection.cu, line=1367) in unknown function, file C:\opencv-2.4.6\modules\gpu\src\cascadeclassifier.cpp, line 173

This function is defined at line 173 of the \opencv-2.4.6\modules\gpu \src\cascadeclassifier.cpp

static void NCVDebugOutputHandler(const std::string &msg) 
    {
        std::cout<<"CUDA debugging error"<<std::endl;
        CV_Error(CV_GpuApiCallError, msg.c_str()); }

Why is this crash happening? How to avoid this crash?

not yet, still looking for an answer

Hi, Face recognition model can be created , saved and reloaded. ( http://docs.opencv.org/modules/contri...)

Can this model be updated with new faces?

Regards,

Thanks. I can't open this file. How can this file be processed?

I could not find the details ofthe trained cascade classifier that comes with OpenCV. Is it possible to get details of the training data set for the frontal face cadcade classifier that comes with OpenCV?

I need to track multiple detected faces from a video stream. Presently, I am using CamShift algorithm to track a detected face from the video stream.

Can it be used to track multiple faces/object?

If not, what can be the alternate to it?

Regards,

// EDIT: Link to crosspost: http://stackoverflow.com/questions/29...

How is the number of windows (integral window) related to he scaling factor factor of the cascade classifier algorithm.

Increasing the scaling factor increases the step size and reduces the number of windows to be evaluated?

Assume, the input image size being evaluated for face detection using cascade classifier is 768x576. How many windows will be there with scale factor 1 and with scale factor 1.1, 1.2, 1.25, 1.30, 1.35, 1.40 etc.

The size of classifier is 14x28.

http://docs.opencv.org/trunk/doc/tutorials/introduction/linux_install/linux_install.html I followed the above tutorial. I do not have x86/build/java folder. Anybody, any clue plz

Dear all,

I am using OpenCV in Windows platform and everything works fine. I am facing decoding error after porting my OpenCV program in Ubuntu 12.04. Below is my setup: Operating System: Ubuntu 12.04 OpenCV version: 2.4.9 Gstreamer is installed

My video streams are H.264 encoded. I am getting the following error:

[mov,mp4,m4a,3gp,3g2,mj2 @ 0x1a22120] error reading header: -12 [mov,mp4,m4a,3gp,3g2,mj2 @ 0x1c66800] error reading header: -12

Does anybody faced this issue? What is the solution to it?

Regards, Tariq

I am using OpenCV 2.4.9 and getting the same error as reported in this thread. What can be the reason for this?

Is cv::gpu::stream support for cascadeclassifier_gpu planned for next OpenCV releases? Can someone help me with OpenCV timeline?

Regards

Is there a way to flush allocated GPU resources before starting a new function call within the same program execution.

I am processing images from videos on GPU. I want to clean up allocated GPU resources after completing one video stream and before starting the processing of the next video stream.

Regards,

I am running 2 detection threads on GPU. My program is crashing in the following lines occasionally.

OpenCV Error: Gpu API call (NCV Assertion Failed: cudaError_t=30, file=C:/opencv-2.4.6/modules/gpu/src/nvidia/NCVHaarObjectDetection.cu, line=1206) in unknown function, file C:\opencv-2.4.6\modules\gpu\src\cascadeclassifier.cpp, line 173

OpenCV Error: Gpu API call (NCV Assertion Failed: cudaError_t=30, file=C:/opencv-.4.6/modules/gpu/src/nvidia/NCVHaarObjectDetection.cu, line=1206) in unknown function, file C:\opencv-2.4.6\modules\gpu\src\cascadeclassifier.cpp, line 173

Error Code: -217

Error Code: -217

Error Description: C:\Users\Abdullah\Downloads\opencv-2.4.6\modules\gpu\src\casc adeclassifier.cpp:173: error: (-217) NCV Assertion Failed: cudaError_t=30, file= C:/opencv-2.4.6/modules/gpu/src/nvidia/NCVHaarObjectDetection.cu, line=1206

Error DescrOpenCV Error: Gpu API call (unknown error) in unknown function, file C:\opencv-2.4.6\modules\core\src\gpumat.cpp, line 1415

Does anybody have idea about it? What can be the reason for it.

I was reading at some nVidia forum that multiple CPU threads execute in the same CUDA context. Could someone please confirm it with reference?

Does cascadeclassifier_gpu support cv::gpu::stream? Could some one please confirm it?

How can I use cv::gpu::stream with cascadclassifier_gpu?

Regards

I can specify NVCC profiler options with CUDA files, with .cu extension, in my project. I don't have any CUDA files in my OpenCV based project. Where shall I specify the NVCC compiler flagfs in this case?

Yes, I compiled the library with CUDA support by myself. Is there any special setting in your visual studio project for enabling the display of kernel launches in Visual profiler?

Thanks for your answers Mathieu. Visual profiler was not helpful with OpenCV GPU module, as this module is already built and included as library in my application. I have no access to CUDA code while executing my application using OpenCV GPU module functions.

Is there a utility that can show GPU usage (workload, memory, #of thread, etc) while an application, using OpenCV GPU module, is running on the GPU?

It will be very helpful in the context of OpenCV GPU module

Regards

Thanks for your comment.

I explained my setup in this thread (http://answers.opencv.org/question/34609/profiling-multi-threaded-opencv-gpu-module/).

I posted the detailed error list in this thread (http://answers.opencv.org/question/34576/opencv-gpu-module-exceptions-from-its-cuda/)

Please go through these and hope the above links would provide you more details to answer my question.

Regards,

How can I determine GPU resource usage from OpenCV GPU module and especially the number of threads used during my program execution?

I am using Tesla K20C for executing my experiments and using GPUTech-Z utility to monitor resource usage.. During the program execution, my GPU is 15-20% loaded and it is consuming about 4% of GPU memory

I tried using Nsight Visual Studio edition, however, it shows only calls to opencv_gpu246d.dll and not to individual CUDA function calls from the library.

I need to debug my cv::gpu module. The OpenCV code is crashing with some errors like CudaLaunchFailure and some errors are like "unknown error". I need to debug the CUDA code from this module. In the profiling results, I can only see one call to Opencv_246gpud.dll.

How can I view the calls made to CUDA kernels from insdie OpenCV GPU module?

I built OpenCV 2,4,6 with CUDA support and I can run my programs on GPU using OpenCV-GPU module.

I want to use cv::gpu::VideoReader_GPU class for directly reading video frames into GPU memory.

I am getting above exception for this line in my code:

VideoReader_GPU flDeviceVideoReader(VIDEO_FILE_NAME);

OpenCV Error: The function/feature is not implemented (The called functionality is disabled for current build or platform) in unknown function, file C:/Users/Ab dullah/Downloads/opencv-2.4.6/modules/gpu/src/precomp.hpp, line 133

What can be the reason for it?

One of the causes for R6010 Abort() is missing DLLs/other resources required by your executable. Hoe this might help

Dear all,

What are best practices/advice for running a multi-threaded application using OpenCV GPU module.

I want to analyse mutliple video streams on GPU in parallel. At the moment, I am creating a separate thread for analysing each file in parallel.

My application works for 2 threads. When I try to run more than 2 threads, I start getting exceptions

Can anybody share his/her experience of developing a multi-threaded application using OpenCV (GPU module) + CUDA? What are the best practices? What to avoid?

What are alternates to using pthreads for multi-threading on GPU (I am unable to use cv::gpu::stream)?

NOTE: It does not appear to be a resource issue. My application is only using 12% of CPU processing and 11% of the GPU meory

Regards,

Thanks @StevenS, It worked.

Dear all,

I am trying to run a multi-threaded application using OpenCV GPU module. The application is developed in Visual Studio 2012 professionl + CUDA 5.5 + OpenCV 2.4.6 and uses pthreads for multi-threading. I built OpenCV 2.4.6 for this environment.

My application executes with two threads and there are no exceptions. I start getting a number of CUDA related exceptions with 3 or more that three threads.

OpenCV Error: Gpu API call (NCV Assertion Failed: cudaError_t=4, file=C:/opencv-2.4.6/modules/gpu/src/nvidia/NCVHaarObjectDetection.cu, line=1306) in unknown function, file C:\opencv-2.4.6\modules\gpu\src\cascadeclassifier.cpp, line 173

OpenCV Error: Gpu API call (unspecified launch failure) in unknown function, file C:\opencv-2.4.6\modules\core\src\gpumat.cpp, line 1196

OpenCV Error: Gpu API call (NCV Assertion Failed: cudaError_t=4, file=C:/opencv-2.4.6/modules/gpu/src/nvidia/NCVHaarObjectDetection.cu, line=1306) in unknown function, file C:\opencv-2.4.6\modules\gpu\src\cascadeclassifier.cpp, line 173

OpenCV Error: Gpu API call (NCV Assertion Failed: cudaError_t=4, file=C:/opencv-2.4.6/modules/gpu/src/nvidia/NCVHaarObjectDetection.cu, line=1523) in unknown function, file C:\opencv-2.4.6\modules\gpu\src\cascadeclassifier.cpp, line 173

OpenCV Error: Gpu API call (unspecified launch failure) in unknown function, file C:\opencv-2.4.6\modules\core\src\gpumat.cpp, line 1415

OpenCV Error: Gpu API call (NCV Assertion Failed: cudaError_t=4, file=C:/opencv-2.4.6/modules/gpu/src/nvidia/NCVHaarObjectDetection.cu, line=1523) in unknown function, file C:\opencv-2.4.6\modules\gpu\src\cascadeclassifier.cpp, line 173

OpenCV Error: Gpu API call (NCV Assertion Failed: retcode=2, file=C:/opencv-2.4.6/modules/gpu/src/nvidia/NCVHaarObjectDetection.cu, line=1768) in unknown function, file C:\opencv-2.4.6\modules\gpu\src\cascadeclassifier.cpp, line 173

OpenCV Error: Gpu API call (NCV Assertion Failed: retcode=5, file=C:/opencv-2.4.6/modules/gpu/src/nvidia/NCVHaarObjectDetection.cu, line=1575) in unknown function, file C:\opencv-2.4.6\modules\gpu\src\cascadeclassifier.cpp, line 173

cudaError_t=4 = cudaErrorLaunchFailure

Any help is appreciated.

Regards,

I am getting the following error during my program execution: NCV Assertion Failed cudaError_t=30 in file NCVHaarObjectDetection.cu at line 1208. I searched a bit and found the explanation for it as under. "This error return is deprecated as of CUDA 3.1. Device emulation mode was removed with the CUDA 3.1 release."

How can I avoid this exception?

Regards,

My program is crashing with following error message:

OpenCV Error: Assertion failed (!empty()) in unknown function, file C:\Opencv-2.4.6\modules\core\src\gpumat.cpp, line 1508

When I traced it. It is referring to the second line of this function from gpumat.cpp file.

GpuMat& cv::gpu::GpuMat::setTo(Scalar s, const GpuMat& mask)
    {
        CV_Assert(mask.empty() || mask.type() == CV_8UC1);
        CV_DbgAssert(!empty());

        gpuFuncTable()->setTo(*this, s, mask);

         return *this;
    }

I added try-catch block around my code.

VideoCapture cap;
cap.open(file);
if (!cap.isOpened())
{
    cerr << "can not open camera or video file" << endl;
    return -1;
}
GpuMat d_frame;
cap >> frame;
try
{
    d_frame.upload(frame);
}
catch(cv::Exception &excep)
{
    std::cout<<excep.what()<<std::endl;
}

However, I still get the same unhandled exception. Any help is appreciated

This is the link for the question [http://answers.opencv.org/question/34476/lbp-for-vehicle-detection/]. Hope to get answer now.

My question is related to this thread by @StevenPuttemans. I need to train LBP classifier for vehicle detection from video streams and answer to the following question would be of great help for me.

1) How much time does it take to train a classifier using LBP for detecting cars?

2) What was the ratio of positive and negative images?

3) Which dataset did you use to train your LBP classifier?

4) How long does it take to train the classifier?

5) Were you able to make a comparison of HaarCascade Classifier and LBP?

6) Is it possible to share the classifier?

@StevenS, I need to train LBP classifier for vehicle detection from video streams and answer to the following question would be of great help for me.

1) How much time does it take to train a classifier using LBP for detecting cars?

2) What was the ratio of positive and negative images?

3) Which dataset did you use to train your LBP classifier?

4) How long does it take to train the classifier?

5) Were you able to make a comparison of HaarCascade Classifier and LBP?

6) Is it possible to share the classifier?

@isarandi, Video is recorded as MP4.

@berak, Video is recorded as MP4 and there are no audio/subtitles in the video.

I am trying to run a multithreaded application in OpenCV using its GPU module. The application is working fine on CPU. I keep the same logice and change the image processing part from CPU to GPU module of OpenCV. My applciation crashes for more than 2 threads in GPU module. Most forums relate it to "out of memory" error. This is not the case for my application as only 495MB out 1GB GPU RAM is being used.

I have described my error in this thread as well.

Some forums suggest using CUDA streams in multi-threaded version of the applicaion for running multiple threads in the application. Can someone provide some example for it?

NOTE: It does not appear to be a resource issue. My application is only using 12% of CPU processing and 11% of the GPU meory

I am running a multithreaded application on GPU using pthreads. My error is not related to memory error (as far as I can see)

I create a separate thread for processing a video file. The code does not crash for 2 threads and performs the job. As soon as I start three threads, my code crashes for 3rd thread at the following line:

OpenCV Error: Gpu API call (unknown error) in unknown function, file C:\Users\Ab dullah\Downloads\opencv-2.4.6\modules\core\src\gpumat.cpp, line 1415 C:\Users\Abdullah\Downloads\opencv-2.4.6\modules\core\src\gpumat.cpp:1415: error : (-217) unknown error

void mallocPitch(void** devPtr, size_t* step, size_t width, size_t height) const
    {
        cudaSafeCall( cudaMallocPitch(devPtr, step, width, height) );
    }

Does anybody have any idea of this crashing behavior?

Dear all,

I am unclear on the use of these two. I updated from OpenCV2.4.3 to OpencCV 2.4.6 and build it for CUDA 5.5 My GPU version of the program in not detecting any objects. Whereas, the same classifier works perfectly on the CPU version

Regards,