Abstract
AffectNet is one of the most popular resources for facial expression recognition (FER) on relatively unconstrained in-the-wild images. Given that images were annotated by only one annotator with limited consistency checks on the data, however, label quality and consistency may be limited. Here, we take a similar approach to a study that re-labeled another, smaller dataset (FER2013) with crowd-based annotations, and report results from a re-labeling and re-annotation of a subset of difficult AffectNet faces with 13 people on both expression label, and valence and arousal ratings. Our results show that human labels overall have medium to good consistency, whereas human ratings especially for valence are in excellent agreement. Importantly, however, crowd-based labels are significantly shifting towards neutral and happy categories and crowd-based affective ratings form a consistent pattern different from the original ratings. ResNets fully trained on the original AffectNet dataset do not predict human voting patterns, but when weakly-trained do so much better, particularly for valence. Our results have important ramifications for label quality in affective computing.
Original language | English |
---|---|
Title of host publication | Pattern Recognition - 6th Asian Conference, ACPR 2021, Revised Selected Papers |
Editors | Christian Wallraven, Qingshan Liu, Hajime Nagahara |
Publisher | Springer Science and Business Media Deutschland GmbH |
Pages | 518-531 |
Number of pages | 14 |
ISBN (Print) | 9783031024436 |
DOIs | |
Publication status | Published - 2022 |
Event | 6th Asian Conference on Pattern Recognition, ACPR 2021 - Virtual, Online Duration: 2021 Nov 9 → 2021 Nov 12 |
Publication series
Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
---|---|
Volume | 13189 LNCS |
ISSN (Print) | 0302-9743 |
ISSN (Electronic) | 1611-3349 |
Conference
Conference | 6th Asian Conference on Pattern Recognition, ACPR 2021 |
---|---|
City | Virtual, Online |
Period | 21/11/9 → 21/11/12 |
Bibliographical note
Funding Information:Acknowledgments. This work was supported by Institute of Information Communications Technology Planning Evaluation (IITP; No. 2019-0-00079, Department of Artificial Intelligence, Korea University) and National Research Foundation of Korea (NRF; NRF-2017M3C7A1041824) grant funded by the Korean government (MSIT).
Publisher Copyright:
© 2022, Springer Nature Switzerland AG.
Keywords
- AffectNet
- Affective computing
- Crowd annotation
- Facial expression recognition
ASJC Scopus subject areas
- Theoretical Computer Science
- General Computer Science