Intelligent waste segregation system

This project demonstrates waste detection using a YOLOv8 (You Only Look Once) object detection model. It identifies recyclable, non-recyclable, and hazardous waste items in a webcam stream.

Our datasets used to train: https://universe.roboflow.com/ai-project-i3wje/waste-detection-vqkjo/model/3

Colab: https://colab.research.google.com/drive/1dHv5QUuz2NkkgzeKBoO4DLAhLg9mOrzv?usp=sharing

Live: https://intelligent-waste-segregation-system.streamlit.app

Setup

Clone the Repository:

git clone https://github.com/boss4848/waste-detection.git
cd waste-detection

Install Dependencies:

pip install -r requirements.txt

Run the Application

streamlit run app.py

Open your web browser and navigate to the provided URL (usually http://localhost:8501). You will see the Waste Detection app.

Project Structure

• app.py: Main application file containing Streamlit code.
• helper.py: Helper functions for waste detection using the YOLO model.
• settings.py: Configuration settings, including the path to the YOLO model and waste types.
• train.py: To train the model

Classifying Waste Items

• RECYCLABLE=['cardboard_box','can','plastic_bottle_cap','plastic_bottle','reuseable_paper']
• NON_RECYCLABLE=['plastic_bag','scrap_paper','stick','plastic_cup','snack_bag','plastic_box','straw','plastic_cup_lid','scrap_plastic','cardboard_bowl','plastic_cultery']
• HAZARDOUS=['battery','chemical_spray_can','chemical_plastic_bottle','chemical_plastic_gallon','light_bulb','paint_bucket']

Screenshots

References

• Streamlit Documentation
• YOLO Documentation