♻️ SmartSort - IoT Smart Waste Segregation Bin

A smart, automated waste management system that uses Computer Vision and IoT to segregate waste into Biodegradable and Non-Biodegradable categories. Powered by a Raspberry Pi 4 and a custom TensorFlow Lite model.

📋 Project Overview

This project solves the problem of improper waste disposal at the source. The bin operates autonomously using a "Sense-Think-Act" pipeline:

1. Sense: An ultrasonic sensor detects when waste is thrown.
2. Think: A camera captures an image, and a TFLite model classifies it (Bio vs. Non-Bio).
3. Act: A servo motor tilts a flap to direct the waste into the correct compartment.

🛠️ Hardware Requirements

• Raspberry Pi 4 (4GB or 8GB RAM recommended)
• Raspberry Pi Camera Module/USB Webcam
• HC-SR04 Ultrasonic Sensor
• SG90 Micro Servo Motor
• External Power Supply for the Servo
• Jumper wires & Breadboard
• Cardboard/Acrylic chassis for the bin

⚙️ Circuit Diagram

The servo motor is powered externally to prevent voltage drops on the Raspberry Pi.

Wiring Pinout:

Component	Pin	Raspberry Pi Pin
Ultrasonic Trig	Trig	GPIO 23
Ultrasonic Echo	Echo	GPIO 24
Servo Signal	Signal	GPIO 18
USB Camera	USB	Any USB Port

Note: The Servo VCC/GND are connected to the external battery pack, sharing a common ground with the Pi.

💻 Software & Dependencies

The project runs on Raspberry Pi OS (RasPiOS).

Prerequisites

• Python 3.7+
• TensorFlow Lite Runtime
• OpenCV
• GPIO Zero

🚀 How to Run

1. SSH into your Raspberry Pi:

   ssh pi@<YOUR_PI_IP_ADDRESS>

2. Navigate to the project directory:

   cd Smart-Waste-Bin

3. Add the scripts:

   # Transfer the Python script
   scp testing.py pi@<YOUR_PI_IP>:/home/pi/Smart-Waste-Bin/
   
   # Transfer the Model file
   scp waste_classifier_model_quantized.tflite pi@<YOUR_PI_IP>:/home/pi/Smart-Waste-Bin/

4. Run the script:

   python3 testing.py

🔄 How the Prototype Runs

Once the script is running, the system follows this autonomous workflow:

1. Initialization: The script initializes the camera and moves the servo motor to the Neutral (Center) position.
2. Waiting for Input: The Ultrasonic Sensor continuously monitors for objects. The terminal will display "Waiting...".
3. Detection: When you bring an object (waste) close to the sensor (within 20cm), the system triggers.
4. Capture & Classification: The camera wakes up, captures a single frame, and sends it to the TFLite model. The prediction (Bio/Non-Bio) is printed in the terminal.
5. Actuation:
   • If Biodegradable: The servo tilts the flap to the Left.
   • If Non-Biodegradable: The servo tilts the flap to the Right.
6. Reset: The system holds the position for 3 seconds to allow the waste to fall, then resets the flap to Neutral and waits 5 seconds before scanning again.

🧠 Model Training

The model was trained using a custom dataset of household waste.

• Classes: Biodegradable, Non-Biodegradable
• Algorithm: SSD MobileNet V2 / YOLO (Specify which one you used)
• Format: Converted to .tflite for edge optimization.

👨‍💻 Contributors

• Shubham Gupta
• Vikas Prajapati
• Sangeet Sangwan
• Prabhkirat Kaur