2016-09-27 03:28:54 -0600 | commented question | reading a samsung recorded h.264 video Thanks for the suggestion. It seems that the video is not corrupted. I have updated my question with the details. |
2016-09-27 03:25:40 -0600 | received badge | ● Editor (source) |
2016-09-26 10:12:55 -0600 | asked a question | reading a samsung recorded h.264 video Hi, I have an issue with reading h264 encoded video recorded by my samsung S5 mini phone camera in OpenCV 3 (on my PC). I get errors like this: and lot of visual artifacts on most of the frames. The strange think is, that vlc and mplayer can play the video OK and converting frames to images with ffmpeg works nice as well. Any suggestions how to fix that? update: As Tetragramm suggested, I have tried to play the video in vlc and look at the statistics. I get no corruptions, but (probably) depending on my current cpu load I get either 0 or quite a lot Lost frames in video section. The visual artifacts are the same as in OpenCV. So it seems that the video file is not corrupted. The question now is how to read it with OpenCV. I don't need to process it in realtime, so is there a way to read it as slow as necessary not to get lost frames? From the VLC codec info about the video: |
2016-07-11 05:32:47 -0600 | received badge | ● Nice Question (source) |
2016-07-09 18:40:52 -0600 | received badge | ● Student (source) |
2016-07-09 18:31:58 -0600 | asked a question | cuda::Filter thread safety Hi, I have found an OpenCV bug regarding cuda::Filters recently. (https://github.com/opencv/opencv/issu...). I would like to fix the bug, but I don't know how to do that. Basically I see two options (as indicated in the comment in the issue tracker). Which one to use? Is there any authority to decide? I don't want to fix the bug in a way that wouldn't be good for the rest of opencv users... thanks! |