<![CDATA[The study established a triathlon bicycle eligibility classification framework based on YoloV11 and YoloV11-OBB, informed by the World Triathlon Competition Regulations 2025, as a solution for the ineffectiveness and subjectivity inherent in manual assessment. The methodology used is Research and Development (R&D), which includes training of deep learning algorithms using 427 side view images and 100 top view images, in addition to evaluating the accuracy of geometric measurements across five different types of bicycles. Findings showed that the model achieved 100% visual classification accuracy as determined by the Confusion Matrix and was able to produce measurements of macro dimensions and component placement with an average margin of error of 1.5-3.3 cm. However, handlebar angle measurements remain prone to perspective distortion, resulting in an average error of 5.21°. In conclusion, the system deserves to be considered as an effective initial screening instrument in the bicycle inspection process, although with the caveat that manual validation is still necessary in cases approaching the regulatory threshold. Keywords: Computer Vision, Oriented Bounding Box, Bicycle Regulations, Triathlon Bike Check, YOLOv11]]>
Copyrights © 2026
