Waste Objects Segregation Using Deep Reinforcement Learning with Deep Q Networks

Effective waste classification is critical in addressing the rising environmental pollution and waste volume.Conventional sorting methods are labor-intensive and error-prone, particularly with the increasing diversity of waste materials.This study presents an innovative approach using deep reinforcement learning for waste object detection and classification to automate waste management processes.The proposed system aims to boost operational efficiency, enhance resource recovery, and reduce waste going to landfills by leveraging deep reinforcement learning.The Deep Q Network model proposed achieved an accuracy of approximately 73%.By employing DQN, an advanced reinforcement learning algorithm, the system ensures improved waste object image classification to handle complex tasks with distributional characteristics.This study can be further extended to design and develop an autonomous waste sorting system.