Real-Time Waste Classification and Localization with YOLOv3

Waste management for solids as is in the current state-of- the-practice, cannot keep up with the waste generation rates of countries, and the currently employed methods of managing solid waste have adverse effects on the environment. In this paper, we propose an alternate approach to how we currently manage solid waste by classification of waste by its type with the use of object-detection systems. We propose an object-detection model that frames object-detection as a regression problem, and it is made to recognize and classify waste, which it does with 89.22% confidence, this helps streamline the process of solid waste management.