低功耗计算机视觉：现状、挑战、机遇
计算机视觉与模式识别
2019-04-17 v1 人工智能
性能
摘要
近年来计算机视觉取得了令人瞩目的进展。同时，手机已成为数百万人的主要计算平台。除手机外，许多自主系统依赖视觉数据做决策，其中部分系统能量有限（如亦称无人机的无人 aerial 飞行器和移动机器人）。这些系统依赖电池，故能效至关重要。本文有两个主要目的：（1）考察图像中物体检测低功耗解决方案的最新水平。自 2015 年起，IEEE 年度国际低功耗图像识别挑战赛（LPIRC）持续举办以识别最节能的计算机视觉解决方案。本文总结 2018 年获奖方案。（2）提出低功耗计算机视觉的研究方向及机遇。
引用
@article{arxiv.1904.07714,
title = {Low-Power Computer Vision: Status, Challenges, Opportunities},
author = {Sergei Alyamkin and Matthew Ardi and Alexander C. Berg and Achille Brighton and Bo Chen and Yiran Chen and Hsin-Pai Cheng and Zichen Fan and Chen Feng and Bo Fu and Kent Gauen and Abhinav Goel and Alexander Goncharenko and Xuyang Guo and Soonhoi Ha and Andrew Howard and Xiao Hu and Yuanjun Huang and Donghyun Kang and Jaeyoun Kim and Jong Gook Ko and Alexander Kondratyev and Junhyeok Lee and Seungjae Lee and Suwoong Lee and Zichao Li and Zhiyu Liang and Juzheng Liu and Xin Liu and Yang Lu and Yung-Hsiang Lu and Deeptanshu Malik and Hong Hanh Nguyen and Eunbyung Park and Denis Repin and Liang Shen and Tao Sheng and Fei Sun and David Svitov and George K. Thiruvathukal and Baiwu Zhang and Jingchi Zhang and Xiaopeng Zhang and Shaojie Zhuo},
journal= {arXiv preprint arXiv:1904.07714},
year = {2019}
}
备注
Preprint, Accepted by IEEE Journal on Emerging and Selected Topics in Circuits and Systems. arXiv admin note: substantial text overlap with arXiv:1810.01732