We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.CV

Change to browse by:

cs

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Computer Vision and Pattern Recognition

Title: Semi-Supervised Disparity Estimation with Deep Feature Reconstruction

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.
Comments: Women in Computer Vision workshop CVPR 2021
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Cite as: arXiv:2106.00318 [cs.CV]
  (or arXiv:2106.00318v1 [cs.CV] for this version)

Submission history

From: Julia Guerrero-Viu [view email]
[v1] Tue, 1 Jun 2021 08:48:38 GMT (2605kb,D)

Link back to: arXiv, form interface, contact.