Unsupervised Domain Adaptation for Depth Prediction from Images

被引:50
|
作者
Tonioni, Alessio [1 ]
Poggi, Matteo [1 ]
Mattoccia, Stefano [1 ]
Di Stefano, Luigi [1 ]
机构
[1] Univ Bologna, Dept Comp Sci & Engn, BO-40126 Bologna, Italy
关键词
Training; Reliability; Estimation; Loss measurement; Computer architecture; Prediction algorithms; Deep learning; depth estimation; unsupervised learning; self-supervised learning; domain adaptation; COST AGGREGATION; STEREO; ACCURATE;
D O I
10.1109/TPAMI.2019.2940948
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
State-of-the-art approaches to infer dense depth measurements from images rely on CNNs trained end-to-end on a vast amount of data. However, these approaches suffer a drastic drop in accuracy when dealing with environments much different in appearance and/or context from those observed at training time. This domain shift issue is usually addressed by fine-tuning on smaller sets of images from the target domain annotated with depth labels. Unfortunately, relying on such supervised labeling is seldom feasible in most practical settings. Therefore, we propose an unsupervised domain adaptation technique which does not require groundtruth labels. Our method relies only on image pairs and leverages on classical stereo algorithms to produce disparity measurements alongside with confidence estimators to assess upon their reliability. We propose to fine-tune both depth-from-stereo as well as depth-from-mono architectures by a novel confidence-guided loss function that handles the measured disparities as noisy labels weighted according to the estimated confidence. Extensive experimental results based on standard datasets and evaluation protocols prove that our technique can address effectively the domain shift issue with both stereo and monocular depth prediction architectures and outperforms other state-of-the-art unsupervised loss functions that may be alternatively deployed to pursue domain adaptation.
引用
收藏
页码:2396 / 2409
页数:14
相关论文
共 50 条
  • [1] Unsupervised Domain Adaptation of Object Detectors: A Survey
    Oza, Poojan
    Sindagi, Vishwanath A.
    Vibashan, V. S.
    Patel, Vishal M.
    IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2024, 46 (06) : 4018 - 4040
  • [2] Contradistinguisher: A Vapnik's Imperative to Unsupervised Domain Adaptation
    Balgi, Sourabh
    Dukkipati, Ambedkar
    IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2022, 44 (09) : 4730 - 4747
  • [3] Learning Domain Invariant Features for Unsupervised Indoor Depth Estimation Adaptation
    Zhang, Jiehua
    Li, Liang
    Yan, Chenggang
    Wang, Zhan
    Xu, Changliang
    Zhang, Jiyong
    Chen, Chuqiao
    ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND APPLICATIONS, 2024, 20 (09)
  • [4] Depth Guidance Unsupervised Domain Adaptation for Semantic Segmentation
    Lu J.
    Shi J.
    Zhu H.
    Sun Y.
    Cheng Z.
    Jisuanji Fuzhu Sheji Yu Tuxingxue Xuebao/Journal of Computer-Aided Design and Computer Graphics, 2024, 36 (01): : 133 - 141
  • [5] Distributionally robust unsupervised domain adaptation
    Wang, Yibin
    Wang, Haifeng
    JOURNAL OF COMPUTATIONAL AND APPLIED MATHEMATICS, 2023, 436
  • [6] CUDA: Contradistinguisher for Unsupervised Domain Adaptation
    Balgi, Sourabh
    Dukkipati, Ambedkar
    2019 19TH IEEE INTERNATIONAL CONFERENCE ON DATA MINING (ICDM 2019), 2019, : 21 - 30
  • [7] Joint Clustering and Discriminative Feature Alignment for Unsupervised Domain Adaptation
    Deng, Wanxia
    Liao, Qing
    Zhao, Lingjun
    Guo, Deke
    Kuang, Gangyao
    Hu, Dewen
    Liu, Li
    IEEE TRANSACTIONS ON IMAGE PROCESSING, 2021, 30 : 7842 - 7855
  • [8] An Unsupervised Monocular Image Depth Prediction Algorithm Based on Multiple Loss Deep Learning
    Tang, Xiaojiao
    Chen, Lifang
    IEEE ACCESS, 2019, 7 : 162405 - 162414
  • [9] Masked GAN for Unsupervised Depth and Pose Prediction With Scale Consistency
    Zhao, Chaoqiang
    Yen, Gary G.
    Sun, Qiyu
    Zhang, Chongzhen
    Tang, Yang
    IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, 2021, 32 (12) : 5392 - 5403
  • [10] Unsupervised Domain Adaptation of Deep Networks for ToF Depth Refinement
    Agresti, Gianluca
    Schaefer, Henrik
    Sartor, Piergiorgio
    Incesu, Yalcin
    Zanuttigh, Pietro
    IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2022, 44 (12) : 9195 - 9208