<![CDATA[Single-use plastic waste, particularly from beverage bottles, remains a significant contributor to the increasing volume of waste in Indonesia. The limited use of reusable beverage containers underscores the urgent need for technological innovations that can support efficient waste segregation. Addressing this issue, the present study proposes a computer vision-based image classification system designed to automatically distinguish between reusable and non-reusable drinking containers. This research adopts a quantitative experimental approach, employing the MobileNetV2 architecture through transfer learning techniques. The model was trained with augmented and normalized datasets to enhance its generalization across diverse image inputs. Evaluation results demonstrate strong classification performance, achieving 96% accuracy, 99% precision (for tumblers), 95% recall, and a 97% F1-score. These outcomes indicate the effectiveness of MobileNetV2 in identifying visual patterns between container types and its potential for deployment in image-driven waste management systems.]]>
Copyrights © 2025
