search

blakeblackshear / frigate

NVR with realtime local object detection for IP cameras
https://frigate.video
MIT License
15.97k stars 1.5k forks source link

Frigate+ submission re-classification of all objects in frame #11529

Closed mjdyson closed 1 month ago

mjdyson commented 1 month ago

Describe what you are trying to accomplish and why in non technical terms I want to reduce the time spent labelling objects within a frigate+ submission so that I can spend more time submitting more photos.

Describe the solution you'd like The current workflow only identifies a single item in the image - I presume this is because it takes the bounding box from the original motion tile (snapshot?) and then overlays it on top of the bigger video frame that it came from. Within this image, there may be multiple objects that are untagged that needs to be bounded and labelled. This starts to get a bit tedious for objects that might be fairly static (cars, plants etc).

It would be great if there was a pipeline that could send a bit more through the local classifier. Perhaps, the whole frame looking for new items, or more pragmatically, an auto-labelling of a new bounding box.

NickM-27 commented 1 month ago

This is a duplicate of https://github.com/blakeblackshear/frigate/issues/8229