Industrial-Grade AI System For Automated Plastic Sorter

This project addresses the pressing issue of plastic waste through an industrial-grade AI sorting system designed for municipal solid waste plastics. Utilizing a dataset of over 33,000 images collected with the UHV Blue Sorter, it captures diverse conditions to simulate real-world scenarios. Advanced deep learning models, including ResNet for classification and YOLOv5 for precise segmentation, ensure accurate sorting, even for challenging materials like dark or deformed plastics. Analytical tools like Grad-CAM and t-SNE offer insights into model performance, guiding improvements in data diversity and feature extraction. By blending cutting-edge AI with industrial adaptability, this scalable and cost-effective system enhances recycling efficiency, contributing to sustainability and reducing environmental impact.