PENERAPAN DEEP LEARNING DENGAN ALGORITMA CNN UNTUK KLASIFIKASI SAMPAH DAUR ULANG DAN B3 (BAHAN BERBAHAYA DAN BERACUN) MENGGUNAKAN ARSITEKTUR RESNET50

Waste management remains a critical issue, especially in distinguishing recyclable waste from hazardous waste (B3). This research aims to develop an automated classification system for eight waste categories using a deep learning-based approach with the ResNet50 architecture. The dataset comprises 2,641 images, divided into four recyclable and four hazardous waste classes. The study was conducted under two scenarios: without and with data augmentation. Results indicate that the model without augmentation performed better, achieving an accuracy of 96%, compared to 94% with augmentation. Evaluation was conducted using precision, recall, and F1-score metrics. Error analysis revealed misclassifications caused by visual similarities between certain image classes. In conclusion, the ResNet50 architecture effectively classifies waste images, although further improvements in data augmentation and dataset diversity are necessary to enhance classification accuracy.