lubo1994/mv-extractor repository overview

⁠Motion Vector Extractor

The motion vector extractor⁠ is a tool to extract frames, motion vectors, frame types and timestamps from H.264 and MPEG-4 Part 2 encoded videos. The tool provides a single class, which serves as a replacement for OpenCV's VideoCapture⁠ and can be used to read and decode video frames from a H.264 or MPEG-4 Part 2 encoded video stream/file.

This Docker image is based on the manylinux_2_28⁠ image and serves two purposes:

1. It contains all dependencies to run the motion vector extractor and its test suite.
2. It functions as build environment for building the Python package⁠ of the motion vector extraction for all supported Python versions.

⁠Tags with respective Dockerfile links

• v1.1.0, latest⁠
• v1.0.6⁠
• v1.0.5⁠
• v1.0.4⁠
• v1.0.3⁠
• v1.0.2⁠
• v1.0.1⁠
• v1.0.0⁠

Images tagged with dev and buildcache are intermediate artefacts generated by the CI and should not be used directly.

⁠Usage

Pull and run the motion vector extractor with

docker run lubo1994/mv-extractor:latest extract_mvs -h

Map a video file into the container and extract motion vectors (replace <videofile.mp4> with the actual filename)

docker run -v ./<videofile.mp4>:/home/video_cap/<videofile.mp4> lubo1994/mv-extractor:latest extract_mvs <videofile.mp4> --verbose

If you want to use the graphical preview, you have to supply additional arguments to the docker run command

docker run -it --ipc=host --env="DISPLAY" -v ./<videofile.mp4>:/home/video_cap/<videofile.mp4> -v /tmp/.X11-unix:/tmp/.X11-unix:rw lubo1994/mv-extractor:latest extract_mvs <videofile.mp4> --preview

For more details on the usage see the project homepage⁠.

⁠About

This software is written by Lukas Bommes⁠ and licensed under the MIT License⁠.

If you use the project for academic research please cite

@INPROCEEDINGS{9248145,
  author={L. {Bommes} and X. {Lin} and J. {Zhou}},
  booktitle={2020 15th IEEE Conference on Industrial Electronics and Applications (ICIEA)}, 
  title={MVmed: Fast Multi-Object Tracking in the Compressed Domain}, 
  year={2020},
  volume={},
  number={},
  pages={1419-1424},
  doi={10.1109/ICIEA48937.2020.9248145}}