Deep affect prediction in-the-wild: aff-wild database and challenge, deep architectures, and beyond
- Submitting institution
-
University of Greenwich
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 29422
- Type
- D - Journal article
- DOI
-
10.1007/s11263-019-01158-4
- Title of journal
- International Journal of Computer Vision
- Article number
- -
- First page
- 907
- Volume
- 127
- Issue
- 6-7
- ISSN
- 0920-5691
- Open access status
- Compliant
- Month of publication
- February
- Year of publication
- 2019
- 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
-
7
- Research group(s)
-
-
- Citation count
- 15
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- Developing large in-the-wild databases has been the main step towards developing state-of-the-art DL-based affect prediction methods. Our Aff-Wild database and AffWildNet are publicly available and widely presented to the research Community (Workshop/Challenge, CVPR-2017; invited talks: EmotiW-Challenge, ICMI-2017; HBU-Workshop, ECCV-2018). AffWildNet was basis for British Patent No.1909300.4 filed by FaceSoft, 2019; it was also provided to Sentido Enterprises, Finland. Aff-Wild was extended (2019) to Aff-Wild2, the largest in-the-wild database, annotated w.r.t. all three affect representations (valence-arousal, facial-expressions, action-units). Aff-Wild2 was used in ABAW Competition/Workshop, IEEE-FG2020, with 62 participating teams. Multi-task AffWildNet provided state-of-the-art performance in predicting all affect representations (BMVC-2019).
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -
