Design and optimization of multiple heads detection for embedded system
Kuei‐Chung Chang, Po-Kai Liu · 2017
Smart human tracking systems based on surveillance camera are very popular recently. For example, retailers and museums use head counting to analyze the consumer statistics. This paper proposed an approach to detect multiple heads, which can be applied in smart human tracking system. The computing resource of this kind of applications is so high that it is not applicable in embedded platforms. So, this paper also proposes a parallel design to enhance the performance of the approach such that it can be more applicable in embedded platforms. Experimental results show the accuracy of head detection is 85.66%. The optimized approach speeds up about 45% in quad-core embedded system.