Store results to file/stdout as JSON or other structured format
Description of Problem & Solution
I am looking at using dvr-scan CLI to detect intervals of high movement in a security/dashcam scenario, and encode the videos more effectively if they are still motion. I intend to use a tdarr (or maybe an alternative) pipeline to constantly process input files, as well as clean up older, insignificant detections. I also would like to build bounding-box overlays without modifying the original files.
Proposed Implementation: When movement is detected, and a bounding box is calculated, build a metadata file with metadata of the detected event(s). The metadata should contain events, as well as the bounding box keyframes (ideally in a unit-less coordinate system), which can then be stored and overlaid at a later date.
Additionally, provide frame metrics, motion score, and possibly an event score to determine how noticeable the event is, allowing for the filtering out of insignificant detections at a later date.
This is a good idea, I'll consider this for the next release.