Design of an Intelligent Recycling System for Renewable Resources in Smart Campuses Based on Visual Detection

With the continuous refinement of national waste classification policies, issues surrounding waste sorting on campuses are also being optimized. To enhance the efficiency of campus waste classification and address the significant time consumption in front-end collection, this paper presents the design of an intelligent classification trash bin based on the Jetson Xavier NX. Utilizing the Jetson Xavier NX development board as the primary control core, the system implements the YOLOv5+DeepSORT dual object detection algorithm for realtime detection of collected waste. It connects to an ATmega2560 microcontroller and an ESP32 IoT board, featuring a self-designed waste classification device, trash bin, and sorting box. A serial display provides real-time classification results and plays promotional videos on waste sorting. When the sorting box reaches capacity, the system autonomously replaces the garbage bag. Through the construction of a prototype and multiple experiments, it was determined that under various lighting conditions, the classification accuracy for different types of waste exceeds 81%, with an average sorting time of 1.3 seconds. Additionally, the system can automatically replace the garbage bag when full. Experimental results indicate that this design not only enhances the accuracy of waste classification but also offers a more userfriendly and efficient solution for campus waste management.