Deep Learning based UAV Assisted Distributed Platform for Crowd Source Management

Person localization and crowd sourcing in aerial Images is a challenging problem and of great interest in many applications such as crowd safety and security management, people management in disasters, and person movement in an epidemic disease situation. However, the Aerial images have an issue of cluttered background, low-resolution quality, and various environment and lighting conditions, which increase complexity levels of localization and recognition. In this paper, we propose a deep learning-based approach on a Raspberry Pi-powered UAV platform. for the acquisition and localization of crowds considering the different orientations of the UAV and heterogeneous environmental factors. We focus on implementing object location and recognition with various constraints. We collected a large local environment dataset for better accuracy. We compared the performance of Haar-Cascade, template matching, YOLO, and convolutional neural network. The aforementioned results show that YOLO performs better for crowd detection and localization.