<![CDATA[In the current digital era, the development of computer vision technology has played a crucial role in various fields, including security and surveillance. This study proposes the application of the YOLOv8 (You Only Look Once version 8) object detection model to identify violations of clothing etiquette in the UIN Suska environment. The proposed approach involves collecting a dataset of 5195 accurately labeled images of clothing and training it using the YOLOv8 architecture, enabling real-time object detection at high speed. The research results indicate that the use of YOLOv8 can recognize and differentiate clothing that adheres to the dress code with good accuracy in the campus context, achieving an mAP of 0.819 at epoch 100 and an F1-Score of 0.79 with a dataset split of 87% training, 8% validation, and 5% testing. By employing appropriate evaluation metrics such as recall, precision, and F1-score, the model's performance can be comprehensively measured to produce optimal results. This research provides a crucial foundation for the development of the YOLOv8 object detection system in maintaining dress code compliance in the campus environment. The study includes the implementation of YOLOv8 on a clothing dataset that reflects the variations commonly found at UIN Suska. Performance evaluation is conducted by comparing detection results with labels provided by humans as ground truth. The use of this model is expected to assist authorities and security staff at UIN Suska in identifying clothing etiquette violations effectively, strengthening supervision, and supporting the creation of a comfortable campus environment in line with institutional values]]>
Copyrights © 2024
