Deep Learning Based Intelligent Waste Identification System

Abstract

Garbage disposal and management have been one of the pertinent issues in every part of the world. Keeping the city clean is the collective responsibility of every citizen. Instead of blaming civic bodies for being waste vigilant, citizens should work with civic bodies to locate and report garbage in a visually informative and effective manner. In order to facilitate this, a novel CNN based garbage management system has been proposed in this work. In addition, an Android application has been developed to help citizens identify, locate, and report litter spots to maintain cleanliness. The user captures the image of the garbage pile. The CNN model identifies the garbage pile. If it is a garbage pile, the location is captured by the app, bundled with the image, and then uploaded into Google Sheets in JSON format via APIs. The app reads the JSON file and updates the location of litter across all the remaining users. The proposed model has obtained an accuracy of 91.23% in garbage identification.