AMFD-YOLO: An Adaptive Multi-scale Feature Modulation Network for Small Object Detection in Airport Surface Surveillance

Abstract Accurate detection of airport surface targets, including aircraft, tow tractors, catering trucks, refueling trucks, baggage carts, and ground staff, is essential for intelligent surface surveillance and automated ground operation management. However, detection remains challenging because of complex and time-varying backgrounds, large target-scale variation, a high proportion of small objects, and frequent mutual occlusion. To address these issues, we propose AMFD-YOLO based on the YOLO11 framework. First, a Multiscale Occlusion-aware Attention Module (MOAM) is embedded in the detection head to enhance responses to occluded targets while suppressing background interference. Second, a C3k2\_FDM module incorporating the Feature Distribution Modulation (FDM) activation function is introduced to improve feature extraction under complex lighting and weather conditions. In addition, a P2 detection layer enhances the perception of tiny targets. Experiments on the Airport Surface Surveillance (ASS) dataset show that AMFD-YOLO improves precision, recall, mAP0.5, and mAP0.5-0.95 by 7.5\%, 23.2\%, 23.6\%, and 12.5\%, respectively, over baseline YOLO11. Ablation studies further verify the contribution of each component. These results demonstrate the effectiveness of AMFD-YOLO for airport surface target detection, particularly under adverse environmental conditions and visually crowded apron and taxiway scenes, and its potential for intelligent airport surveillance.