Stream
WebSocket + backpressure loop
Real-Time Object Detection Dashboard
Webcam / demo-feed object-detection dashboard with FastAPI WebSocket streaming, YOLO-ready detector abstraction, canvas bounding-box overlays, and live throughput / latency / confidence stats.
Python 3.11FastAPIWebSocketsReact + ViteTypeScriptYOLOv8 (optional)OpenCV (optional)Docker Compose
Detector
Mock or YOLOv8 adapter
Project Overview
Production-shaped CV dashboard: stream URL + FPS controls, live detection stage with overlays, stats cards (active, confidence, throughput, latency), confidence timeline, and event stream. Mock detector by default; swap in YOLOv8 with weights.
Challenge
Computer vision portfolios usually live in notebooks — hard for hiring teams to actually run, and disconnected from real product surfaces.
Solution
Wrapped a detector interface behind a FastAPI WebSocket with a frame-processing backpressure loop, and a React dashboard that overlays bounding boxes and aggregates live metrics.
Results
Self-contained demo that runs in mock mode with zero setup, then upgrades to real YOLOv8 when weights are dropped in.