<![CDATA[This research aims to develop a computer vision-based Indonesian rupiah currency denomination detection system that can provide automatic voice output as accessibility support for the visually impaired. The model was developed using the YOLOv11 algorithm, which was customized to recognize seven nominal classes of 2022 issue banknotes. The public dataset was used as training, validation, and test data, which was then processed thru transformations and augmentations to improve model generalization. Training was conducted using a controlled configuration with AdamW optimization and overfitting prevention strategies. Performance evaluation was conducted using the metrics of accuracy, precision, recall, F1-score, mean Average Precision, confusion matrix, and ROC-AUC. The research results show that the model achieved very high performance with an accuracy of 99.64% and mAP of 0.9935, indicating consistent identification capabilities for currency denominations across all classes. The simple OpenCV-based implementation and voice conversion using gTTS prove that the model can operate in real-time and provide direct audio feedback. This finding indicates that YOLOv11 is effective for Indonesian rupiah recognition and has the potential for further development in accessibility applications for the visually impaired.]]>
Copyrights © 2026
