I have video material containing frames with a rectangular area of uniform color interspersed with other content. I would like to identify these frames, with the intent of removing them. The technique discussed here looks promising, as does the ffmpeg blackdetect filter on its own. However, I need to be able to specify the color of the rectangle for each video. It is not always black.
Identify a logo should be part of the CV. My suggestion is to use the idea of checking the logo from png (https://ai-facets.org/robust-logo-detection-with-opencv/) and modify the code to read a frame from a video (https://opencv-python-tutroals.readthedocs.io/en/latest/py_tutorials/py_gui/py_video_display/py_video_display.html)
Hope this help