Counterfactual Depth from a Single RGB Image

被引:2
|
作者
Issaranon, Theerasit [1 ]
Zou, Chuhang [1 ]
Forsyth, David [1 ]
机构
[1] Univ Illinois, Urbana, IL 61801 USA
关键词
D O I
10.1109/ICCVW.2019.00268
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
We describe a method that predicts, from a single RGB image, a depth map that describes the scene when a masked object is removed - we call this "counterfactual depth" that models hidden scene geometry together with the observations. Our method works for the same reason that scene completion works: the spatial structure of objects is simple. But we offer a much higher resolution representation of space than current scene completion methods, as we operate at pixel-level precision and do not rely on a voxel representation. Furthermore, we do not require RGBD inputs. Our method uses a standard encoder-decoder architecture, and with a decoder modified to accept an object mask. We describe a small evaluation dataset that we have collected, which allows inference about what factors affect reconstruction most strongly. Using this dataset, we show that our depth predictions for masked objects are better than other baselines.
引用
收藏
页码:2129 / 2138
页数:10
相关论文
共 50 条
  • [1] Attention Unet++ for lightweight depth estimation from sparse depth samples and a single RGB image
    Tao Zhao
    Shuguo Pan
    Wang Gao
    Chao Sheng
    Yingchun Sun
    Jiansheng Wei
    [J]. The Visual Computer, 2022, 38 : 1619 - 1630
  • [2] Deep Depth Completion of a Single RGB-D Image
    Zhang, Yinda
    Funkhouser, Thomas
    [J]. 2018 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2018, : 175 - 185
  • [3] Contextualized CNN for Scene-Aware Depth Estimation From Single RGB Image
    Song, Wenfeng
    Li, Shuai
    Liu, Ji
    Hao, Aimin
    Zhao, Qinping
    Qin, Hong
    [J]. IEEE TRANSACTIONS ON MULTIMEDIA, 2020, 22 (05) : 1220 - 1233
  • [4] Stereo-augmented Depth Completion from a Single RGB-LiDAR image
    Choi, Keunhoon
    Jeong, Somi
    Kim, Youngjung
    Sohn, Kwanghoon
    [J]. 2021 IEEE INTERNATIONAL CONFERENCE ON ROBOTICS AND AUTOMATION (ICRA 2021), 2021, : 13641 - 13647
  • [5] Attention Unet plus plus for lightweight depth estimation from sparse depth samples and a single RGB image
    Zhao, Tao
    Pan, Shuguo
    Gao, Wang
    Sheng, Chao
    Sun, Yingchun
    Wei, Jiansheng
    [J]. VISUAL COMPUTER, 2022, 38 (05): : 1619 - 1630
  • [6] Two-Stage Deep Regression Enhanced Depth Estimation From a Single RGB Image
    Sun, Jianyuan
    Wang, Zidong
    Yu, Hui
    Zhang, Shu
    Dong, Junyu
    Gao, Pengxiang
    [J]. IEEE TRANSACTIONS ON EMERGING TOPICS IN COMPUTING, 2022, 10 (02) : 719 - 727
  • [7] Depth estimation from a single RGB image using target foreground and background scene variations
    Alphonse, P. J. A.
    Sriharsha, K., V
    [J]. COMPUTERS & ELECTRICAL ENGINEERING, 2021, 94
  • [8] Depth Estimation and Semantic Segmentation from a Single RGB Image Using a Hybrid Convolutional Neural Network
    Lin, Xiao
    Sanchez-Escobedo, Dalila
    Casas, Josep R.
    Pardas, Montse
    [J]. SENSORS, 2019, 19 (08)
  • [9] Depth Estimation From a Single RGB Image Using Fine-Tuned Generative Adversarial Network
    Ul Islam, Naeem
    Park, Jaebyung
    [J]. IEEE ACCESS, 2021, 9 (09): : 32781 - 32794
  • [10] Spectral Reflectance Recovery From a Single RGB Image
    Fu, Ying
    Zheng, Yongrong
    Zhang, Lin
    Huang, Hua
    [J]. IEEE TRANSACTIONS ON COMPUTATIONAL IMAGING, 2018, 4 (03): : 382 - 394