Computer Vision Based Drone Detection Using Mask R-CNN
Auwalu Saleh Mubarak, M. Vubangsi, Fadi M. Al-Turjman, Zubaida Said Ameen, Abdulmunaim Saleh Mahfudh, Sinem Alturjman · 2022 International Conference on Artificial Intelligence in Everything (AIE) · 2022
Security of private and military areas is very important, drones are becoming more available day by day even though they are used for different purposes. In no-fly zones like metros, airports, private properties and so on, unauthorized drones can be detected using computer vision models, in this study Mask-RCNN with two different backbones (ResNet-50 and MobileNet) was employed to detect drones. It was observed that the Mask-RCNN with ResNet-50 as backbone achieves mAP of 0.965, mAP-50 of 0.991, mAP-75 of 0.958 and recall of 0.961.