Real-Time Flying Object Detection and Counting Using Deep Learning

This paper presents a general model for real-time flying object detection that can be used for transfer learning and further research, as well as a refined model that achieves state-of-the-art results. to detect flying objects. We do this by training a dataset containing 40 different types of flying objects with fine-tuned data and handling some practically unlikely cases such as objects at long distances due to Camera placement and flying objects are often in the sky, or objects are often very similar such as AG600 and US-2 seaplanes, making identification difficult. The article uses the YOLO model with the latest versions: YOLOv8-S, YOLOv8-N, YOLOv9-T, and YOLOv10-T. Finally, we achieved feasible and outstanding results with the parameters mAP50 92.8% and mAP50-95 87.3% of the best YOLOv8-N model when compared with YOLOv8-S, YOLOv9-T and YOLOv10-N. In addition, the paper also achieved the best results for the YOLOv10-N model when counting the number of objects in the frame customized according to given requirements to ensure control of the number of objects in the frame to control the number of entries and exits. This applies to airport control or aviation security safety.