Automatic Waste Segregator Using Image Classification

Abstract

Negligent and improper waste segregation has led to a lot of environmental issues like global warming and we and the upcoming generations are the victims to it. Improper segregation causes production of waste that cannot be reutilized which makes itself hazardous as the dumping of plastic. Hence Waste Segregation is of utmost necessity. A fully automated waste segregation system would be beneficial as it would enable us to implement a fully sealed garbage sorting space, so the chances of pollution of garbage reduces. As it is automated, the odour problem from waste can be handled. It would also mean that the workers' health in the waste management industry would be prone to lesser risks. Final products procured from the waste segregation process can be refined into useful resources or disposed without generation of any pollutant. This paper proposes an "Automatic Waste Segregator" (AWS) which can be used by organizations handling large number of people and hence dealing with a lot of day to day waste. The waste classifier is built using an Image Classification Algorithm, trained and tested on the Waste images dataset. The classifier segregates the waste into three categories namely, dry, wet and plastic waste. The classifier is then deployed on a Raspberry Pi board. The trash will be thrown into a dustbin which is internally divided into three bins. All three bins are covered by plates which form the common base where connected to it) will then detect the type of waste and the corresponding plate opens and the trash falls into the right bin. The AWS is connected to a network hence security is of utmost importance. We propose multi factor authentication to access any data or IOT device related to the AWS. In this paper we also deal with next generation firewall for protection of the network the project works on and also ensuring hardware outage doesn’t result in insecure state of IOT device.