A Sustainable Approach to Waste Detection and Classification using YOLO: A Review

The increasing volume of discarded materials worldwide requires advanced waste management strategies to address environmental pollution and promote sustainability. The long-term environmental impacts of waste highlight the importance of proper identification, sorting, and utilization of different waste categories. This review paper synthesises recent advancements in approaches for waste segregation, focusing on computer vision (CV) and deep learning algorithms, particularly various variants of the YOLO (You Only Look Once) model. There are different models with efficacy in identifying waste and its categories. But, still challenges remain with these models. This paper presents a methodology for managing waste sustainably by using classification and feedback systems, which helps in improving the accuracy of the model. Along with studying different papers, this work shows the methods, datasets and performance in each paper. The analysis focuses on these technologies mainly in terms of accuracy, processing time so as to promote sustainable waste management in real life.

Keywords— Deep Learning models, computer vision (CV), waste detection and classification, YOLO, transfer learning, waste management, object segmentation.