<![CDATA[This community service activity aims to develop an automated waste sorting system based on Deep Learning using the YOLOv11 architecture for real-time object detection. The system is designed to identify six categories of waste, namely cardboard, glass, metal, paper, plastic, and trash through direct camera input. The dataset used was obtained from Kaggle, consisting of 2,467 images which were subsequently expanded to 10,108 images through data augmentation techniques to improve model generalization and reduce the risk of overfitting. The training process employed an 80% training set and a 20% validation set, with optimization using the Stochastic Gradient Descent (SGD) algorithm, momentum, and cosine learning rate scheduling over 100 epochs. Model performance was evaluated using the standard object detection metric, Mean Average Precision (mAP). The results show that the model achieved an mAP@0.5 of 95.8% and an mAP@0.5:0.95 of 82.4%, with an average inference speed of 45 frames per second (FPS) on standard hardware. These findings indicate that the system is capable of performing waste classification quickly, accurately, and efficiently, supporting the implementation of smart waste management in urban environments, particularly in collaboration with waste management partners in Depok.]]>
Copyrights © 2026
