摘要
介绍了利用深度学习从图像或视频中恢复三维结构、进行深度估计和实现视觉传感器实时定位方面的研究与应用;对深度学习的研究概况进行了介绍;深入分析和比较了有无监督情况下具有代表性的深度学习算法和系统;对近年来深度学习方面的研究热点进行了讨论,并进行了总结和展望。
Research and application of deep learning in recovery of three-dimensional structure from image or video, depth estimation, and real-time localization of visual sensor are introduced. Research progress of deep learning is overviewed. According to whether there is supervision, some representative deep learning algorithms and systems are introduced individualiy with deep analysis and comparison. Finally, the research spots on deep learning in recent years are discussed, conclusions are presented, and some research tendencies are discussed.
作者
鲍振强
李艾华
崔智高
袁梦
Bao Zhenqiang;Li Aihua;Cui Zhigao;Yuan Meng(Rocket Force University of Engineering, Xi'an, Shaanxi 710025, China)
出处
《激光与光电子学进展》
CSCD
北大核心
2018年第5期62-70,共9页
Laser & Optoelectronics Progress
基金
国家自然科学基金(61501470)
陕西省重点研发计划(2017GY-075)
关键词
视觉光学
三维感知
视觉定位
深度估计
深度学习
研究进展
visual optics
three-dimensional perception
vision localization
depth estimation
deep learning
research progress