Cascaded Collaborative Regression for Robust Facial Landmark Detection Trained Using a Mixture of Synthetic and Real Images With Dynamic Weighting
- Submitting institution
-
The University of Surrey
- Unit of assessment
- 12 - Engineering
- Output identifier
- 9000586_2
- Type
- D - Journal article
- DOI
-
10.1109/TIP.2015.2446944
- Title of journal
- IEEE Transactions on Image Processing
- Article number
- -
- First page
- 3425
- Volume
- 24
- Issue
- 11
- ISSN
- 1057-7149
- Open access status
- Out of scope for open access requirements
- Month of publication
- -
- Year of publication
- 2015
- 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)
-
-
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- Introducing a new paradigm in automatic facial landmark detection by augmenting real faces with synthetically generated images, the approach makes innovative use of Surrey’s 3D face model to generate the additional annotated 2D face image data required. The collaborative methodology of training achieves much better performance in terms of landmarking accuracy and has provided the enabling technology for our internationally leading low/cross resolution face matching solution (10.1109/TBIOM.2020.3007356), of interest to the security industry (Digital Barriers) and government agencies (Home Office, dstl). It helped secure a £2.4m EPSRC project in retrieval from multimodal archives with BBC, (Surrey, Ulster and Cambridge (EP/V002856/1)).
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -