Waste Manager

An automated waste sorting system that combines computer vision and Arduino control to classify waste and route it physically in real time.

Overview

The project uses a camera with YOLO for object detection, a Gemini vision model for material classification, and an Arduino Uno to control doors and a conveyor belt based on the classification result. So in brief the system detects an object, classifies its material, and activates motors and doors to guide it to the correct path.

Labels

• 0: Carton
• 1: Plastic
• 2: Other

Pins

• Pin 2: Responsible for returning the left and right doors to their normal state (closed).
• Pin 3: Opens the left door to route carton waste after classification.
• Pin 4: Opens the right door to route plastic waste after classification.
• Pin 7: Runs the conveyor belt, moving the waste item through the system during sorting.

Pipeline

1. The camera runs continuously and monitors the scene in real time.
2. YOLO performs object detection on the live video feed.
3. When an object is detected and the cooldown period has passed, a frame is captured.
4. The captured image is saved to the images folder with a timestamp.
5. The saved image is sent to the Gemini vision model for waste classification.
6. The model returns a single label: 0 (carton), 1 (plastic), or 2 (other).
7. The label is sent from Python to the Arduino via serial communication.
8. The Arduino starts the sorting cycle by activating Pin 2 to reset the system.
9. Based on the received label:
   • Label 0 (Carton): Opens the left door (Pin 3) and runs the belt (Pin 7).
   • Label 1 (Plastic): Opens the right door (Pin 4) and runs the belt (Pin 7).
   • Label 2 (Other): Keeps both doors closed and only runs the belt (Pin 7).
10. After the belt finishes running, all pins are reset and the system waits for the next object.