The Use of Convolutional Neural Networks (CNN) for the Identification of Recycling and Garbage

Abstract

Using deep learning and image processing algorithms to better identify waste in public settings, this study aims to enhance municipal trash collection. Better waste management techniques and the development of smart cities are both aided by this study. Two CNNs were trained using the AlexNet architecture; one was trained to detect garbage in images, and the other to distinguish between landfill rubbish and recyclables. To demonstrate the idea, the two-stage CNN system was first trained and evaluated on the industry-standard TrashNet indoor picture dataset, where it obtained impressive results. The authors then trained and evaluated the system using outside photographs captured in the actual setting of use. The first convolutional neural network (CNN) was able to obtain a preliminary accuracy of 93.6% in identifying garbage and non-trash objects on a picture database of various types of rubbish using the outdoor image dataset. Accuracy levels ranged from 89.7 percent to 93.4 percent, with a total of 92.2 percent, achieved by training a second CNN to differentiate between landfill waste and recyclables. The eventual aim is to include this garbage recognition system that is based on image processing into a smart garbage can robot that can identify and gather garbage in its immediate vicinity by taking pictures in real-time. Cullowhee, North Carolina's Yanjun Yan School of Engineering and Technology Dear Sir/Madam, unwelcoming to strangers, and particularly harmful to children and the elderly. In industrialized nations, inhabitants face a major hygienic, aesthetic, and environmental protection issue with garbage and refuse that goes uncollected along roads and other locations. Worldwide, "unsafe water supply, sanitation and hygiene" are responsible for an estimated 842,000 fatalities annually, according to the World Health Organization [1]. Under five-year-olds make up 361,000 of them, with the majority residing in nations with low per capita income. Trash collection costs a pretty penny in both wealthy and developing nations, but automatic methods will cut that cost in half while also boosting public health. For instance, according to CBSNewYork [2], the annual cost of garbage collection in New York City is $300 million.