[Libav-user] Subject: Inquiry Regarding Obtaining Motion Vectors in FFMPEG

[mxlai at mail.ustc.edu.cn]

Dear FFMPEG Community,

I hope this email finds you well. My name is Minxin Lai, a graduate student working on system optimization. I am currently exploring the FFMPEG tool for my work and have a specific question regarding motion vectors.

I recently came across a project on GitHub, specifically the mv-extractor tool (https://github.com/LukasBommes/mv-extractor), and I noticed that there is an issue reported that aligns with a challenge I am facing. The issue can be found at: https://github.com/LukasBommes/mv-extractor/issues/2

The problem I am encountering is similar to what is described in this issue. It seems that the tool decodes the entire video frame, and I am interested in retrieving motion vectors without the need for full frame decoding to reduce CPU usage.

I understand from the issue discussion that there might be a possibility to modify the tool to achieve this goal, but I am not familiar with the necessary changes in the code. As someone who is not an expert in C++, I am reaching out to seek guidance on potential solutions or alternative approaches to address this issue.

If you have any insights on how I could modify the tool or if there are specific resources I should refer to for a better understanding, I would greatly appreciate your assistance. Your expertise in this area would be invaluable for my research.

Thank you for your time and consideration. I look forward to any guidance you can provide.

Best regards,

Minxin Lai

University of Chinese Academy of Sciences


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://ffmpeg.org/pipermail/libav-user/attachments/20231208/3a46ce4b/attachment.htm>