Unsupervised Multi-View Constrained Convolutional Network for Accurate Depth Estimation

doi:10.1109/TIP.2020.2997247

CORC > 自动化研究所 > 中国科学院自动化研究所 > 模式识别国家重点实验室 > 三维可视计算

	Unsupervised Multi-View Constrained Convolutional Network for Accurate Depth Estimation
	Zhang, Yuyang 2,3; Xu, Shibiao 2,3; Wu, Baoyuan 1; Shi, Jian 2,3; Meng, Weiliang 2,3; Zhang, Xiaopeng 2,3
刊名	IEEE TRANSACTIONS ON IMAGE PROCESSING
	2020
卷号	29 页码:7019-7031
关键词	Estimation Training Feature extraction Geometry Computer vision Cameras Unsupervised learning Unsupervised learning DenseDepthNet multi-view geometry constraint depth consistency
ISSN号	1057-7149
DOI	10.1109/TIP.2020.2997247
通讯作者	Xu, Shibiao(shibiao.xu@nlpr.ia.ac.cn) ; Meng, Weiliang(weiliang.meng@ia.ac.cn)
英文摘要	Accurate depth estimation from images is a fundamental problem in computer vision. In this paper, we propose an unsupervised learning based method to predict high-quality depth map from multiple images. A novel multi-view constrained DenseDepthNet is designed for this task. Our DenseDepthNet can effectively leverage both the low-level and high-level features of input images and generate appealing results, especially with sharp details. We employ the public datasets KITTI and Cityscapes for training in an end-to-end unsupervised fashion. A novel depth consistency loss based on multi-view geometry constraint is also applied to the corresponding points across pairwise images, which helps to improve the quality of predicted depth maps significantly. We conduct comprehensive evaluations on our DenseDepthNet and our depth consistency loss function. Experiments validate that our method outperforms the state-of-the-art unsupervised methods and produce comparable results with supervised methods.
资助项目	National Key Research and Development Program of China[2018YFB2100601] ; National Natural Science Foundation of China[61620106003] ; National Natural Science Foundation of China[91646207] ; National Natural Science Foundation of China[61971418] ; National Natural Science Foundation of China[61771026] ; National Natural Science Foundation of China[61972459] ; National Natural Science Foundation of China[61671451] ; National Natural Science Foundation of China[61571046] ; National Natural Science Foundation of China[61561003]
WOS研究方向	Computer Science ; Engineering
语种	英语
出版者	IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
WOS记录号	WOS:000546910100009
资助机构	National Key Research and Development Program of China ; National Natural Science Foundation of China
内容类型	期刊论文
源URL	[http://ir.ia.ac.cn/handle/173211/40092]
专题	模式识别国家重点实验室_三维可视计算
通讯作者	Xu, Shibiao; Meng, Weiliang
作者单位	1.Chinese Univ Hong Kong, Shenzhen 518172, Peoples R China 2.Chinese Acad Sci, Inst Automat, Natl Lab Pattern Recognit, Beijing 100190, Peoples R China 3.Univ Chinese Acad Sci, Sch Artificial Intelligence, Beijing 100049, Peoples R China
推荐引用方式 GB/T 7714	Zhang, Yuyang,Xu, Shibiao,Wu, Baoyuan,et al. Unsupervised Multi-View Constrained Convolutional Network for Accurate Depth Estimation[J]. IEEE TRANSACTIONS ON IMAGE PROCESSING,2020,29:7019-7031.
APA	Zhang, Yuyang,Xu, Shibiao,Wu, Baoyuan,Shi, Jian,Meng, Weiliang,&Zhang, Xiaopeng.(2020).Unsupervised Multi-View Constrained Convolutional Network for Accurate Depth Estimation.IEEE TRANSACTIONS ON IMAGE PROCESSING,29,7019-7031.
MLA	Zhang, Yuyang,et al."Unsupervised Multi-View Constrained Convolutional Network for Accurate Depth Estimation".IEEE TRANSACTIONS ON IMAGE PROCESSING 29(2020):7019-7031.