<![CDATA[This study aims to develop an automatic recognition system for aircraft maintenance tools and flight instruments using the YOLO11 model in an aviation education environment. The dataset was collected from the Avionic Laboratory and consisted of five object classes: Airspeed Indicator, Flight Director, Pictorial Navigation Instrument, Crimping GMT 208, and Crimping GMT 221. The model was trained iteratively through three stages using data augmentation techniques and evaluated using precision, recall, mAP50, mAP50-95, confusion matrix analysis, and live video testing. The best performance was achieved in the third iteration with an mAP50-95 value of 0.85082, precision of 0.96013, recall of 0.99207, and mAP50 of 0.97450. Real-time testing demonstrated that the system performed very well under normal conditions; however, its performance decreased when visual disturbances were introduced. The confusion matrix analysis showed that the model most frequently confused the two crimping tool classes due to their similar visual characteristics. Overall, the results indicate that YOLO11 has strong potential to be implemented as an effective and adaptive learning support tool in aviation vocational education.]]>
Copyrights © 2026
