An Embedded System for Real-Time 3D Human Detection
Title: An Embedded System for Real-Time 3D Human Detection
Authors: Haibin Cai (Computer Science, Loughborough University); Lei Jiang (Computer Science, Loughborough University); Junyi Wang (Computer Science, Loughborough University); Mohamad Saada (Computer Science, Loughborough University); Qinggang Meng (Computer Science, Loughborough University);
Year: 2019
Citation: Cai, H., Jiang, L., Wang, J., Saada, M., Meng, Q., (2019). An Embedded System for Real-Time 3D Human Detection. UK-RAS19 Conference: “Embedded Intelligence: Enabling & Supporting RAS Technologies” Proceedings, 116-117. doi: 10.31256/UKRAS19.31
Abstract:
Recent years have seen great achievements in the design of deep learning network structures and the construction of large benchmark datasets for object detection. However, it still remains a great challenge to achieve real-time performance when adapting to embedded systems with low computational ability. This paper proposes an embedded system for real-time 3D human detection. The system consists of a neural computing stick for the deploy of CNN, an intel RGBD sensor for the 3D sensing and a Raspberry Pi platform. Furthermore, a novel multi-thread based human detection framework is proposed to improve the detection speed. Experimental results show that the system can effectively detection human in real-time performance.