Farshad's profile - activity

I am seeing something strange with GoodFeaturesToTrack() function in C++. When in Debug mode, the function works as expected, returning the number of features I am asking it via maxCorners argument. But when I run it in release mode, the function returns much larger number of features, none of which correlate with the output in debug mode. I have initialized the input image and properly allocated for the number of maxCorners I need. Has anyone seen this behavior before? I am also trying to step into the code but could not find the source file in the bundle I downloaded. I am using OCV 2.4.9 in windows 64 bit.

I convert the original video to RGB outside OpenCV using VLC Media Player.

Additional information possibly relevant to this discussion (extracted from VLC Media Player)

Stream 0 Type: Video Code: 8 bits greyscale (GREY) Resolution: 1280x720 Display Resolution: 1280x720 Frame Rate: 30.096

thanks, Farshad.

Thanks Steven. In doing this, I get segmentation fault in FFMPEG_proxy (CvCapture_FFMPEG_proxy class). Is there any special setting for the FFMPEG decoder I have to implement? I am completely baffled with the complexity of this decoder so any help is greatly appreciated.

Hi, I was wondering if there was a way to read in greyscale video files in OpenCV. I currently have to convert greyscale video into RGB format outside OpenCV, read it in and again convert it to greyscale using cvtColor() function... Is there anyway I can omit these counterintuitive steps and just grab greyscale frames from original video files?

I am trying to read a video file into a vector up to free memory into the system. As the content of the vector is being pulled out by a separate thread additional frames are to be read in by the loop reading the video file. The issue I am running into is that I do not see any means to put the loop on hold until the vector size reduces to certain size. Is there a better way to implement this concurrency?