Implementasi Algoritma CNN Untuk Pemilahan Jenis Sampah Berbasis Android Dengan Metode CRISP-DM

Abstract

Garbage is a global problem, especially in developing countries. People often don't sort the waste properly, causing contaminated waste that can't be recycled. An Android-based application using Convolutional Neural Network (CNN) algorithm can help people sort waste correctly. The application receives input in the form of waste photos taken by users and classifies the waste types using CNN algorithm. The results are displayed to the user with accurate information to dispose of waste into the right trash can. The testing using Transfer Learning CNN with Mobile Net architecture on 5,428 waste datasets resulted in 97.95% precision and 95.18% recall. Using tensorflow Lite library on Android, banana peel waste can be detected with 96% accuracy as organic waste and cardboard waste can be detected with 99% accuracy as inorganic waste