Eagle AI
A smart camera platform that answers two questions in real time: what's in the frame, and who is in the frame. Built during my internship at ABICOM Software.
Overview
Eagle AI turns ordinary camera feeds into a monitored, searchable system: live object detection, face recognition against enrolled identities, camera management, and analytics, all streaming to a web dashboard in real time.
My work spanned the Python/FastAPI detection backend and the React frontend: wiring YOLOv8 detections and InsightFace embeddings into live data that an operator can actually use, frame by frame.
What I built
- Real-time object detection with YOLOv8, stabilized across frames with BoT-SORT multi-object tracking.
- Face recognition using InsightFace embeddings, matched against an enrolled-persons database.
- Live streaming to the browser over WebRTC, with WebSocket channels pushing detection events as they happen.
- Camera management and analytics dashboards in React, Tailwind, and MUI.
- Async MongoDB for high-write telemetry, plus an FFmpeg pipeline for stream ingest.
Next project