Computer Vision for Automated Aircraft Surface Inspection

Research Problem

Aircraft structures are continuously exposed to operational wear and environmental stress throughout their lifetime. Detecting surface defects such as cracks, corrosion, dents, scratches, or missing rivets is therefore essential to ensure aircraft safety and airworthiness. Today, this process still relies heavily on manual visual inspections performed by trained maintenance personnel. While effective, these inspections are time-consuming, labor-intensive, and dependent on human judgment, which may introduce variability in detection performance.

Recent advances in computer vision and artificial intelligence offer new opportunities to support and partially automate aircraft inspection processes. By combining image-based sensing technologies with machine learning algorithms, it becomes possible to automatically detect potential defects across large aircraft surfaces. However, practical deployment requires robust algorithms, reliable data collection methods, and representative datasets captured under realistic inspection conditions.

Contributions

Our research investigates the potential of computer vision techniques for automated aircraft surface inspection, combining algorithm development with experimental validation using real-world inspection data.

Key contributions include:

• Development and evaluation of real-time computer vision algorithms for aircraft defect detection based on modern object detection architectures such as YOLO and RT-DETR
• Collection of a real-world dataset of aircraft surface defects, captured using drone imagery from a preserved aircraft with visible defects located in Korat, Nakhon Ratchasima (Thailand)
• Demonstration of the feasibility of low-cost inspection workflows using drone-based imaging combined with AI models, supporting more efficient aircraft inspection processes

This research contributes toward more efficient, scalable, and data-driven aircraft inspection processes.

Selected Publications

• Advances in Aircraft Skin Defect Detection Using Computer Vision: A Survey and Comparison of YOLOv9 and RT-DETR Performance
  Suvittawat, N., Kurniawan, C., Datephanyawat, J., Tay, J., Liu, Z., Soh, D. W., & Ribeiro, N. A. (2025). Aerospace.
• Aircraft Surface Defect Inspection System Using AI with UAVs Suvittawat, N., & Ribeiro, N. A. (2023). International Conference on Research in Air Transportation (ICRAT).