Bilateral Cyclic Constraint and Adaptive Regularization for Unsupervised Monocular Depth Prediction

被引:62
|
作者
Wong, Alex [1 ]
Soatto, Stefano [1 ]
机构
[1] Univ Calif Los Angeles, Vis Lab, Los Angeles, CA 90095 USA
关键词
IMAGE-RESTORATION;
D O I
10.1109/CVPR.2019.00579
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Supervised learning methods to infer (hypothesize) depth of a scene from a single image require costly per-pixel ground-truth. We follow a geometric approach that exploits abundant stereo imagery to learn a model to hypothesize scene structure without direct supervision. Although we train a network with stereo pairs, we only require a single image at test time to hypothesize disparity or depth. We propose a novel objective function that exploits the bilateral cyclic relationship between the left and right disparities and we introduce an adaptive regularization scheme that allows the network to handle both the co-visible and occluded regions in a stereo pair. This process ultimately produces a model to generate hypotheses for the 3-dimensional structure of the scene as viewed in a single image. When used to generate a single (most probable) estimate of depth, our method outperforms state-of-the-art unsupervised monocular depth prediction methods on the KITTI benchmarks. We show that our method generalizes well by applying our models trained on KITTI to the Make3d dataset.
引用
收藏
页码:5627 / 5636
页数:10
相关论文
共 50 条
  • [1] An Adaptive Unsupervised Learning Framework for Monocular Depth Estimation
    Yang, Delong
    Zhong, Xunyu
    Lin, Lixiong
    Peng, Xiafu
    IEEE ACCESS, 2019, 7 : 148142 - 148151
  • [2] Generative Adversarial Networks for Unsupervised Monocular Depth Prediction
    Aleotti, Filippo
    Tosi, Fabio
    Poggi, Matteo
    Mattoccia, Stefano
    COMPUTER VISION - ECCV 2018 WORKSHOPS, PT I, 2019, 11129 : 337 - 354
  • [3] Adaptive Co-teaching for Unsupervised Monocular Depth Estimation
    Ren, Weisong
    Wang, Lijun
    Piao, Yongri
    Zhang, Miao
    Lu, Huchuan
    Liu, Ting
    COMPUTER VISION - ECCV 2022, PT I, 2022, 13661 : 89 - 105
  • [4] Monocular image depth prediction without depth sensors: An unsupervised learning method
    Chen, Songnan
    Tang, Mengxia
    Kan, Jiangming
    APPLIED SOFT COMPUTING, 2020, 97
  • [5] UnDispNet: Unsupervised Learning for Multi-Stage Monocular Depth Prediction
    Kaushik, Vinay
    Lall, Brejesh
    2019 INTERNATIONAL CONFERENCE ON 3D VISION (3DV 2019), 2019, : 633 - 642
  • [6] An unsupervised monocular image depth prediction algorithm using Fourier domain analysis
    Chen, Lifang
    Tang, Xiaojiao
    IET SIGNAL PROCESSING, 2022, 16 (06) : 619 - 629
  • [7] Unsupervised Depth and Confidence Prediction from Monocular Images using Bayesian Inference
    Bhutani, Vishal
    Vankadart, Madhu
    Jha, Omprakash
    Majumder, Anima
    Kumar, Swagat
    Dutta, Samrat
    2020 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS (IROS), 2020, : 10108 - 10115
  • [8] Unsupervised Monocular Depth Estimation for Monocular Visual SLAM Systems
    Liu, Feng
    Huang, Ming
    Ge, Hongyu
    Tao, Dan
    Gao, Ruipeng
    IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, 2024, 73 : 1 - 13
  • [9] Monocular Depth Estimation Based on Unsupervised Learning
    Liu, Wan
    Sun, Yan
    Wang, XuCheng
    Yang, Lin
    Zheng, Zhenrong
    OPTOELECTRONIC IMAGING AND MULTIMEDIA TECHNOLOGY VI, 2019, 11187
  • [10] Unsupervised Learning of Monocular Depth from Videos
    Gao Haosheng
    Teng Wang
    2019 CHINESE AUTOMATION CONGRESS (CAC2019), 2019, : 3945 - 3950