A Computer Vision-Based Plastic Bottle Classification and Recycling Management System Using Enhanced Yolov8n with Se Block and CBAM

Abstract

This study developed a Prototype CV-Based Plastic Bottle Classification and Recycling Management System using YOLOv8n model integrated with SE Block and CBAM is designed to improve waste segregation and recycling practices within the Mindanao State University – Sultan Naga Dimaporo (MSU-SND) campus. The system utilized a Raspberry Pi 4, Raspberry Pi Camera Module, servo motor, LCD display, and the YOLOv8n object detection model for real-time bottle detection and sorting. The study adopted the Software Development Life Cycle (SDLC) using the Spiral Model, which involved iterative phases of planning, risk analysis, design, development, testing, and evaluation to ensure continuous system improvement. A dataset composed of plastic and non-plastic bottle images underwent preprocessing and augmentation techniques to improve detection accuracy and model robustness. The dataset was divided into training, validation, and testing subsets using a 70:20:10 ratio to ensure reliable model evaluation. Furthermore, the YOLOv8n model, enhanced with Squeeze-and-Excitation (SE) Block and Convolutional Block Attention Module (CBAM), was trained to improve feature extraction and detection performance under varying lighting conditions, object positions, and orientations. The trained model achieved a mAP@0.5 score of 0.97 demonstrating high object detection and classification performance during validation and testing. During operation, the system identified whether the inserted item was plastic or non-plastic and automatically directed it into the appropriate bin using a servo-controlled diverter mechanism while displaying corresponding feedback through the LCD module. The developed system demonstrated effective real-time detection, classification, and automated sorting performance, showing its potential in supporting sustainable waste management and promoting environmental responsibility within the campus community.