FWAVs Object Detection Based on YOLOv8 In Adverse Weather

In recent years, Flapping-wing Micro Aerial Vehicles (FMAVs) have gained increasing attention due to their biomimetic appearance and excellent maneuverability. Equipping them with object detection systems to make them more intelligent is becoming a major trend. This paper proposes an aerial photography small object detection algorithm under adverse weather conditions based on YOLOv8. Using a custom dataset, the concluding mAP50 (mean Average Precision at Intersection over Union threshold of 0.5) reached 0.761, surpassing other datasets. The mAP50 improved from 0.627 to 0.708 compared to model validation on the VisDrone2019 dataset, surpassing other mainstream object detection models.