OPTIMASI IMAGE CLASSIFICATION PADA JENIS SAMPAH DENGAN DATA AUGMENTATION DAN CONVOLUTIONAL NEURAL NETWORK

Abstract

Garbage is useless goods/materials used normally or specifically in production, goods damaged during production or useless materials which mainly come from households. Moreover, inorganic waste is very difficult and takes a longer time to be decomposed by the soil. The lack of public knowledge about the classification of types of waste and how to process it causes a very serious problem in Indonesia. Therefore, this research creates a waste type recognition program using the Convolutional Neural Network (CNN) algorithm, which can be used to detect and recognize objects in an image. CNN is a technique inspired by the way mammals, humans, produce visual perception. CNN is included in the type of deep neural network because of its high network depth and widely applied to imagery. 2 Types of waste classification, namely inorganic waste and organic waste. The implementation of garbage image recognition uses 2 test models, Sequential and on top VGG16 which runs on the Google Collaboratory application, and Keras. After carrying out the Augmentation process, the number of test data in this study was 1489 images on the training data and 182 on the testing data resulting in an evaluation value with an accuracy of 90.97% and a loss value of 0.307 on the Sequential model, and an accuracy value of 97.99% with a loss value of 0.069 on the on top model. VGG16.