Multi-UAV Collaborative Face Recognition for Goods Receiver in Edge-Based Smart Delivery Services

Yi Xu, Fengguang Luan, Jonathan Kua, Haoyu Luo, Zhipeng Wang, Xiao Liu

Abstract: The accurate and efficient recognition of goods receivers is a crucial task for Unmanned Aerial Vehicle (UAV) based smart delivery services in edge computing. Meanwhile, face recognition has become one of the major recognition methods in smart delivery services given its high efficiency and reliability. However, unlike many other application scenarios, the face recognition algorithm’s accuracy and efficiency are often limited due to the unique camera shooting angles by the UAVs flying in the sky. To address such an issue, in this paper, we propose a multi-UAV collaborative face recognition method to enhance both the accuracy and efficiency of face recognition for multiple goods receivers at the same time, named UAVs4FR. Specifically, instead of each UAV running its own face recognition process, multiple UAVs first use face detection models to collaboratively detect and capture pedestrian faces, then transmit these images to the edge server. Afterwards, the edge server uses a face recognition algorithm to perform preliminary recognition of faces. Meanwhile, the face fusion model is also deployed in the edge server to fuse the face images to improve the recognition accuracy. Finally, the edge server utilizes a face recognition algorithm to further recognize target goods receivers using the fused face images and send out the corresponding instructions to the group of UAVs through broadcasting. Comprehensive experiments based on a real-world edge-based smart UAV delivery service system successfully demonstrate the effectiveness of our proposed framework and its better performance compared with other representative methods.