Dynamic Attention-Controlled Cascaded Shape Regression Exploiting Training Data Augmentation and Fuzzy-Set Sample Weighting
- Submitting institution
-
The University of Surrey
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 9018801_3
- Type
- E - Conference contribution
- DOI
-
10.1109/CVPR.2017.392
- Title of conference / published proceedings
- 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)
- First page
- 0
- Volume
- -
- Issue
- -
- ISSN
- -
- Open access status
- -
- Month of publication
- -
- Year of publication
- 2017
- 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
-
-
- Research group(s)
-
-
- Citation count
- 30
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- The key contribution of this paper is a dynamic attention mechanism that improves the fault tolerance of cascaded facial landmark detection algorithms. This method achieves impressive results that are even better than deep-learning-based state-of-the-art approaches. This algorithm ranked 4th on the profile face subset of the Menpo facial landmark localisation challenge, organised in CVPR2017. Many other researchers followed this work by replacing some of the components by using deep neural networks and achieved even better results in unconstrained facial landmark localisation.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -