Waste Segregation using YOLOv8 and Robotics
■ Project Overview
This project addresses the challenge of manual waste segregation in India by introducing an
automated system that leverages computer vision and robotics. Using YOLOv8 for real-time object
detection, the system can classify waste into recyclables, non-recyclables, and other categories.
ByteTrack is integrated for multi-object tracking to ensure continuous monitoring. A Delta robotic
arm with a suction pump is used for pick-and-place operations, enabling efficient physical
segregation of waste. The combination of AI and robotics minimizes human intervention, speeds up
recycling, and improves accuracy in waste management.
■ Tools and Technologies
Technology Purpose
YOLOv8 Real-time waste object detection and classification
ByteTrack Multi-object tracking for continuous monitoring
Dataset 29,231 images of bottles, plastics, cans, tetra packs with augmentation
Delta Arm + Suction Pump Robotic pick and place for physical segregation
Camera Module Captures real-time waste images
Arduino + Servo Motors Controls robotic arm movement
Python Backend development and YOLO integration
React UI User interface for monitoring and manual control
■ Conclusion
The project demonstrates a modern approach to sustainable waste management by combining
deep learning (YOLOv8) and robotics (Delta Arm). With high detection accuracy (98.3% mAP50)
and efficient real-time segregation, this solution addresses the limitations of existing methods like
CNNs and YOLOv3. It provides a scalable, reliable, and eco-friendly approach for automated waste
management in smart cities.
