Implementation Of Machine Learning To Identify Types Of Waste Using CNN Algorithm

Abstract

Waste management remains a significant challenge globally, particularly in Indonesia, where the annual waste generation reached 24.67 million tonnes in 2021, with only 50.43% properly managed. To address the issue of mixed organic and inorganic waste and the lack of public awareness regarding waste separation, this study applied machine learning, specifically the Convolutional Neural Network (CNN) algorithm, to classify waste types. The research aimed to develop an effective automated waste classification model to improve waste management processes. The research involved collecting a dataset of 2,848 images representing six waste categories: glass, cardboard, paper, metal, organic, and plastic. Preprocessing techniques such as cropping, noise reduction with Gaussian filters, and data augmentation were applied to enhance data quality. The dataset was divided into training, validation, and testing subsets in a 70:20:10 ratio. The CNN model employed feature extraction through convolution, activation, and pooling layers, followed by classification using a fully connected layer and a softmax function. Model performance was evaluated using accuracy, precision, recall, and F1-score metrics. The model achieved an overall accuracy of 95%, with an average precision, recall, and F1-score of 0.95 across all classes. These results demonstrate the CNN model’s ability to reliably classify waste types. Compared to previous studies, this research achieved higher accuracy through the use of enhanced preprocessing and CNN optimization. This study highlights the potential of CNN-based models for automated waste classification, contributing to sustainable waste management practices and fostering environmental awareness in the future research.