A review of real-time object detection for sustainable edge deployment in waste sorting sites

Edge computing for real-time object detection is becoming increasingly important in places like waste sorting facilities, where resources and space are limited. Accurate and fast detection and identification of deformable objects in these environments can greatly improve recycling efficiency and overall performance. This review evaluates the performance of various real-time object detection models — both end-to-end and non-end-to-end — on the Nvidia Jetson Orin. It also discusses about dataset acquisition, industrial constraints, and potential strategies for achieving optimized detection, identification and execution time.