TrashCam

AI-Powered Smart Trashcan — Computer Vision + Raspberry Pi + Retro UI

TrashCam is an end-to-end smart waste-monitoring system using computer vision, Raspberry Pi, and a retro-styled dashboard to detect waste, analyze fill levels, track mis-sorting, and provide LED feedback on a real trash bin.

The system pipeline is:

Camera → Raspberry Pi → Laptop (AI + Retro UI) → Pi Disaplay

---

Features

Live Camera Capture

• Intel Navilense Camera
• Streams MJPEG/RTSP over LAN
• Low-latency capture for tracking falling trash in real time

AI Waste Classification

• YOLOv8-World (or custom YOLO model)
• Groups predictions into:
  • recycling
  • organic
  • general waste
• Optional deduping avoids repeat logs for the same item

Retro Dashboard (React UI)

• CRT-style “retro terminal” look
• Displays:
  • Fill level bar
  • Current detected trash
  • Mis-assignment %
  • Event log feed
  • Real-time device status
• Works stand-alone (backend optional initially)

LED Feedback (Raspberry Pi)

• WS2812B or Neopixel LEDs
• Visual feedback for:
  • correct sorting
  • contamination detected
  • bin nearing full
  • system errors / warnings

Logging & Stats

• JSON event logging
• CSV analytics
• Daily usage summaries
• Mis-sort tracking for environment analysis

---

Tech Stack

Frontend

• React
• Retro CRT theme (CSS)
• ShadCN UI
• Live-updating stats & graphs

Backend (Python)

• FastAPI
• OpenCV
• YOLOv8-World inference
• Logging + analytics

Hardware

• Raspberry Pi 5
• Intel Navilense Camear
• WS2812B LED strip
• Trash bin