Computer Vision Utilities

This repo is for keeping my utilties written with python and pytorch. Please see this repository for pytorch project template.

Requirements

• poetry
• python >= 3.8
• pytorch >= 1.0
• pytest
• pytest-mock
• pre-commit (for pre-commit formatting, type check and testing)
• opencv
• opencv-contrib-python = 3.x

Please run poetry install to install the necessary packages.

List

• mixup
• Motion History Image
• optical flow
  • Lucas-Kanade
  • Gunnar Farneback
  • Dual TVL1 (NOTE: This is implemented by opencv and takes a lot of time)
• Image Scrambling
• Bounding box selector (for annotation)

TODO

• While pytest runs without any errors using poetry, it fails in github actions though dependencies are installed with poetry as well. This may be caused by opencv-python or opencv-contrib-python.
• use GPUs for calculating Dual TVL1 optical flow

License

This repository is released under the MIT License

Reference

• James W. Davis and Aaron F. Bobick "The Representation and Recognition of Action Using Temporal Templates" in CVPR 1997
• H. Zhang+ "mixup: BEYOND EMPIRICAL RISK MINIMIZATION" in ICLR2018
• B. D. Lucas and T. Kanade (1981), "An iterative image registration technique with an application to stereo vision" in Proceedings of Imaging Understanding Workshop, pages 121--130
• Gunnar Farneb¨ack, "Two-frame motion estimation based on polynomial expansion", in Scandinavian conference on Image analysis (pp. 363-370). Springer, Berlin, Heidelberg, 2002.