Structured Uncertainty Prediction Networks
- Submitting institution
-
The University of Bath
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 210664210
- Type
- E - Conference contribution
- DOI
-
10.1109/CVPR.2018.00574
- Title of conference / published proceedings
- 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition
- First page
- 5477
- Volume
- -
- Issue
- -
- ISSN
- 1063-6919
- Open access status
- Compliant
- Month of publication
- December
- Year of publication
- 2018
- URL
-
-
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
4
- Research group(s)
-
-
- Citation count
- 2
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- This work provides a practical approach to quantifying long range correlations in error distributions that is missed by nearly all deep generative models that either ignore uncertainty or adopt a per pixel approach that results in poor calibration of errors. The denoising application outperforms approaches specifically trained to denoise and the theoretical contributions open many potential applications of generative models where uncertainty is important e.g. medical imaging (inverse problems) and driverless cars (segmentation and depth estimation). Invited presentations at the Rank Prize Symposium, BMVA Technical Meeting, Computer Vision Overview Symposium, Max Planck Institute for Perceiving Systems and Siemens Research.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -
