I've been working on training YOLOv8 to identify my cats specifically. We've had some issues with them getting on the counters and going outside of the litter box. Previous camera systems to try and find the culprit we not always triggered by the cat's motion. I'd wanted to move to a self hosted camera system anyway so I got an IP camera and modified Frigate NVR to support ultralytics and slapped my custom trained model in. The training has been interesting as I've never done anything with computer vision before this. The hardest part was getting enough pictures and then labeling them. If I retrain it in the future I'd like to use the trained model to identify any cats in the source pictures and then just fix the ones it missed or overzealously marked.