The aim of this review is to summarize the development in the field of AI driven visual waste recognition and disposal coordination. Achieved by sampling a multitude of studies reflecting the transition from traditional techniques to the advancement such as hybrid deep learning integrating object detection, segmentation and real time analysis. The initial attempts used CNN classifiers having limitations in generalization and deployment. Latter improvements utilizing models like YOLO, Mask R-CNN and edge based model gives better accuracy in an controlled environment However, there are persistent challenges like limited dataset diversity, unstable performance in dynamic conditions and lack of real-time integration. It was observed that simultaneously implementing both classification and segmentation are critical to enable spatial precision and content understanding is a critical observation. This review identifies key research gaps and presents future directions for scalable, context-aware AI waste systems.
Parihar et al. (Wed,) studied this question.