Intelligent Waste Sorting System Based on YOLO-Enhanced Deep Learning Models

This research proposes an intelligent waste management system integrating YOLOv11, a state-of-the-art object detection algorithm, with Large Language Models for automated waste classification and reporting. Our system addresses the challenges of urban waste management by combining visual recognition capabilities with textual analysis for comprehensive waste categorization. Experimental results demonstrate that YOLOv11 outperforms previous YOLO variants, achieving 0.931 precision and 0.971 recall in bounding box mode, with mAP@50-95 reaching 0.874. In mask mode, YOLOv11 attains even higher performance with 0.933 precision and 0.978 recall, surpassing both YOLOv5 and YOLOv8 variants. The multimodal approach enables accurate detection and classification while generating detailed disposal recommendations, significantly enhancing waste management efficiency.