AI-Based Real-Time Waste Sorting System for Recycling Plants Using YOLO and Deep Learning Models

Abstract

Modern recycling facilities must have effective waste segregation because of the growing amount of mixed garbage and the ineffectiveness of manual sorting techniques. Manual segregation is timeconsuming, labor-intensive, and puts workers' health at risk. This research describes an AI-based real-time waste sorting system that uses deep learning and YOLO approaches to overcome these issues. The YOLO model concurrently detects and classifies various trash objects in the proposed system, which uses a camera to record live video streams of rubbish on a conveyor belt. The waste that was found is divided into five categories: glass, metal, organic, plastic and e-waste. By preventing duplicate detections, object tracking guarantees precise counting. While Flask offers a web-based interface for real-time viewing of detection and counting results, OpenCV is used for image processing.