Over the past few decades, interest in algorithms for face recognition has been growing rapidly and has even surpassed human-level performance. Despite their accomplishments, their practical integration with a real-time performance-hungry system is not feasible due to high computational costs. So in this paper, we explore the recent, fast, and accurate face recognition system that can be easily integrated with real-time devices, and tested the algorithms on robot hardware platforms to confirm their robustness and speed.
翻译:在过去几十年里,对面部识别算法的兴趣一直在迅速增长,甚至超过了人的性能。 尽管它们取得了成就,但由于计算成本高,它们与实时性能饥饿系统的实际整合并不可行。 因此在本文件中,我们探索了可以很容易与实时设备整合的近期快速准确面部识别系统,并测试了机器人硬件平台上的算法,以确认其坚固性和速度。