FURI | Summer 2025

Real-Time MCU-Based Smart Recycling Bin with Embedded Computer Vision

This study investigates whether a cost-effective microcontroller (MCU)-based system integrated with basic computer vision can classify common recyclable materials (paper, plastic and metal) in real time. A lightweight convolutional neural network (CNN) optimized for the MCU achieved over 80% classification accuracy under varying lighting conditions, demonstrating feasibility for household adoption. By automating sorting at the point of disposal, the system reduces contamination during recycling and enhances efficiency in waste management. Future work will focus on expanding material categories, improving robustness under external environmental factors, and integrating wireless data logging for large-scale deployment and monitoring.