The sharp urbanization with the growing consumption develops the emission of solid waste that gives serious environmental and obstacles on the waste management. Well-separation of waste and recyclable stuffs is an important measure leading to a sustainable waste processing method; nevertheless, hand sorting is time consuming, laborious, and error-prone. This paper alleviates these drawbacks by proposing an automated system of waste identification using deep learning and computer vision methods. The recommended system applies an object detection model that is convolutional neural net to classify and detect the waste materials, including plastic, metal, paper, glass, and non-recyclable trash through visual images. This model examines real-time visuals or video streams to precisely identify types of waste thus making it possible to segregate the waste automatically. The system will run on edge computing platform so that there is low latency and its operation is cost-effective. It has been proven by experiments that the proposed strategy has a high classification accuracy and steady real-time performance. Moreover, the system may be combined with intelligent bins or robotic platforms to be deployed in terms of real time in urban areas. This solution would enable an effective and scaled out smart waste management and recycling usages.
Bhavana et al. (Thu,) studied this question.