Design and Implementation of an AI-Based Automatic Plastic Waste Sorting System

This paper presents the design and implementation of an intelligent plastic waste sorting system that integrates computer vision, robotics, and a graphical user interface (GUI) for real-time supervision. The system employs the YOLOv8 (You Only Look Once) deep learning model for the detection and classification of seven plastic categories: PET, HDPE, single-use plastics, single-layer plastics, multi-layer plastics, squeeze tubes, and UHT boxes. A robotic manipulator (EEZYbotARM Mk2), controlled by an STM32F446RE microcontroller, performs automated pick-and-place operations based on the detection results. A custom PyQt5 GUI provides live video streaming, detection zones and results, category counters, and features such as counter reset, detection filtering, and export of detection history to Excel. The system was validated on a conveyor-based setup, demonstrating accurate detection, reliable robotic sorting and an intuitive user interface. This work addresses the limited development of plastic recycling in emerging countries, highlighting a feasible and efficient solution to improve plastic waste management.