Semi-Supervised Disparity Estimation with Deep Feature Reconstruction

J. Guerrero-Viu, S. Izquierdo, Philipp Schröppel, Thomas Brox

IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Workshops, Jun 2021

Abstract: Despite the success of deep learning in disparity estimation, the domain generalization gap remains an issue. We propose a semi-supervised pipeline that successfully adapts DispNet to a real-world domain by joint supervised training on labeled synthetic data and self-supervised training on unlabeled real data. Furthermore, accounting for the limitations of the widely-used photometric loss, we analyze the impact of deep feature reconstruction as a promising supervisory signal for disparity estimation.

Paper Poster

Images and movies

BibTex reference

@InProceedings{SB21c,
  author       = "J. Guerrero-Viu and S. Izquierdo and P. Schr{\"o}ppel and T. Brox",
  title        = "Semi-Supervised Disparity Estimation with Deep Feature Reconstruction",
  booktitle    = "IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Workshops",
  month        = "Jun",
  year         = "2021",
  keywords     = "disparity estimation, semi-supervised, feature reconstruction",
  url          = "http://lmbweb.informatik.uni-freiburg.de/Publications/2021/SB21c"
}

Other publications in the database