AI Crowd Control Detection System Implemented on FPGA Hardware Development Platform
Chung-Bin Wu, Yuhu Wu, Yi-Yen Lai · 2022
This paper proposes a detection application for crowd control. Detecting objects is mainly to identify pedestrians at close range. The deep learning network uses Yolo-like. And change Yolo-like[1] into a network architecture that only recognizes pedestrians at close distances. The hardware uses Convolution and Detection Layer IP hardware accelerators. And realize the function of Maxpooling and Shortcut on FPGA platform. The FPGA platform is Xilinx Zynq UltraScale+ MPSoC ZCU102 Evaluation Kit. Finally, the FPGA platform uses HDMI to display the recognition results that only detect pedestrians at close distances. Achieve the detection effect of crowd control.