ymmzs / video_stitch Goto Github PK

python video_to_img.py /path/to/video [homography_type]

The 'homography_type' parameter has three options: 'get_homography_manual', 'get_homography_opencv', and 'get_homography_superglue'. The default is 'get_homography_manual'.

'get_homography_manual' calculates the homography matrix using manually annotated matching points. 'get_homography_opencv' calculates the homography matrix by extracting feature points using SuperRetina and obtaining matching point pairs through OpenCV. 'get_homography_superglue' computes the homography matrix using feature points extracted by SuperRetina and matching point pairs obtained via Superglue.

For example,

python video_to_img.py ROP_1.avi get_homography_manual

The result of the Mosaic using the manually labeled feature points to calculate the homography matrix is as follows: