Automated Detection of Atypical Aviation Obstacles from UAV Images Using a YOLO Algorithm.

Area of Science:

• Aerospace Engineering
• Computer Vision
• Geospatial Analysis

Background:

• Increasing urban development near airports creates atypical aviation obstacles that threaten air traffic safety.
• Accurate identification and classification of these obstacles are crucial for maintaining safe airspace.
• The irregular shapes of these obstacles challenge automated height determination methods.

Purpose of the Study:

• To analyze the automated detection of atypical aviation obstacles using the YOLO algorithm.
• To evaluate the accuracy of height determination for these obstacles using Unmanned Aerial Vehicle (UAV) data.
• To enhance aviation safety through improved obstacle detection techniques.

Main Methods:

• Utilized Unmanned Aerial Vehicles (UAVs) for high-resolution data acquisition.
• Applied the YOLO (You Only Look Once) algorithm for automated object detection.
• Analyzed the accuracy of height measurements derived from UAV-based data.

Main Results:

• Demonstrated the capability of the YOLO algorithm in detecting atypical aviation obstacles.
• Assessed the accuracy of automated height determination from UAV-derived spatial data.
• Identified challenges associated with irregular obstacle shapes in automated detection.

Conclusions:

• Automated detection of atypical aviation obstacles using UAVs and YOLO is feasible.
• The study provides insights into the accuracy of height determination for improved aviation safety.
• Further research can refine algorithms for more robust obstacle detection and measurement.