Smart Waste Segregation System using Convolutional Neural Networks

Abstract

As the population increases, waste management has become difficult in today’s world. It is estimated that, until now in India in the year 2022, the total amount of waste generated is 62millon tons, wherein nearly 43 million tons of waste is collected in which about 12million tons is treated and 31million tons is dumped in landfill. Moreover, segregation of waste also is a tough task for the workers. Since the waste may be hazardous and infectious for the human life, the segregation must be carried out without labor. This problem is majorly being faced in developing cities. To overcome this, a smart waste segregation system using CNN is proposed in this paper. Segregation of waste from building itself brings about a large change in waste management. As the waste gets collected in a dumping area, it is identified using Open CV with the help of Pi camera. The captured image of waste is compared with default images using CNN algorithm, and segregation is done using robotic arm. Furthermore, the ultrasonic sensor present within the bin monitors the level of waste; when the waste level reaches the maximum, it alerts the authorized person with the help of GSM module. As the result, this keeps the buildings clean and supports the swatch Bharath mission. Existing waste segregation system use moisture sensor to differentiate between wet and dry waste ,even with small water droplets present on dry or metal waste detects it as wet which is a wrong identification.