AI Accelerator Design to Optimize Power Consumption for Smart Camera

In today’s era of booming artificial intelligent (AI) applications, smart camera products are widely used in many different practical application environments such as observation, monitoring, recognition and identification. Due to the working characteristics, the smart camera is always used continuously for a long time, increasing the energy consumption. Therefore, smart cameras need a solution to optimize power consumption to ensure system performance while employing artificial intelligence (AI) algorithms in image processing. Designing a smart camera system requires careful attention to this aspect. In this paper, a 2-frequency conversion mode method is proposed to improve the energy efficiency of smart camera systems using AI accelerators. Two test scenarios were conducted covering a series of test cases to demonstrate that the energy consumption of the system is significantly improved compared to previous methods.