A computer vision system for detecting and tracking predator-prey dynamics in video. Built on top of YOLOv7 with custom fine-tuning for wildlife scenarios.
Live demo: arvind6599.github.io/PredatorPreyWebsite
What it does
- Detects and classifies animals in video frames in real time
- Tracks movement and infers predator-prey relationships across frames
- Deployed as a web app with a clean interface for uploading and analyzing footage
Stack
- Model: YOLOv7 (fine-tuned on custom dataset)
- Training: PyTorch
- Deployment: GitHub Pages + Python backend