Abstract:
The detection of foreign object debris (FOD) in real-time is crucial for airport safety and security. In this paper, we propose a holistic FOD detection framework that ut...Show MoreMetadata
Abstract:
The detection of foreign object debris (FOD) in real-time is crucial for airport safety and security. In this paper, we propose a holistic FOD detection framework that utilizes both present and past data. A baseline map is constructed first based on frames extracted from clean runway/taxiway videos. During periodic inspections, detection-time frames and aligned baseline images are combined and fed into a modified YOLOv7 network to detect FOD. We explore and compare different configurations of the detection networks to determine the tradeoffs in performance.
Published in: 2023 IEEE 33rd International Workshop on Machine Learning for Signal Processing (MLSP)
Date of Conference: 17-20 September 2023
Date Added to IEEE Xplore: 23 October 2023
ISBN Information: