AiGenics: Smart Waste Classification & Tracking

AiGenics is an end-to-end full-stack IoT computer vision pipeline. It utilizes a custom-trained YOLOv8 Nano model to identify and track waste categories in real-time, pushing aggregated telemetry data to a MERN stack command center for visual analytics.

System Architecture

1. Edge Inference (Python/OpenCV): Captures live webcam feeds and runs a custom YOLO model (trained on the TACO dataset).
2. Object Tracking (ByteTrack): Implements multi-object tracking to assign unique, persistent IDs to individual items, preventing duplicate database entries for a single piece of waste.
3. REST API (Express/Node.js): Receives structured JSON payloads from the edge device.
4. Database (MongoDB): Stores relational data between physical Smart Bins and high-frequency waste logs.
5. Dashboard (React/Vite/Recharts): A responsive, auto-refreshing interface visualizing waste distribution analytics.

Tech Stack

• Computer Vision: PyTorch, Ultralytics YOLOv8n, OpenCV, ByteTrack
• Backend: Node.js, Express.js, Mongoose
• Frontend: React, Vite, Recharts
• Database: MongoDB

Local Setup & Installation

1. The MERN Backend

```bash cd backend npm install

Create a .env file with PORT=5000 and your MONGO_URI

node server.js ```

2. The React Command Center

```bash cd frontend npm install npm run dev ```

3. The Edge AI

Ensure you have Python installed and your virtual environment activated. ```bash pip install -r requirements.txt

Update the BIN_OBJECT_ID in src/inference.py with your MongoDB SmartBin _id

python src/inference.py ```

Engineering Notes

• Duplicate Log Mitigation: Migrated from a brute-force time-based cooldown approach to implementing ByteTrack. This allows the system to establish object permanence, reliably tracking items across frames and ensuring a 1:1 ratio of physical objects to database entries.