Sample images of C-EXPR-DB of subjects of different ethnicities, genders and age groups. The in-the-wild nature of the database can be seen, with high variations in head and body poses, gestures, lightning and illumination conditions, backgrounds.
Research in automatic analysis of facial expressions mainly focuses on recognising the seven basic ones. However, compound expressions are more diverse and represent the complexity and subtlety of our daily affective displays more accurately. Limited research has been conducted for compound expression recognition, because only a few databases exist, which are small, lab controlled, imbalanced and static. In this work, I present an in-the-wild A/V database, C-EXPR-DB, consisting of 400 videos annotated in terms of 13 compound expressions, valence-arousal emotion descriptors, action units, speech, facial landmarks and attributes.
If you are an academic, (i.e., a person with a permanent position at a university, e.g. a professor, but not a Post-Doc/PhD/PG/UG student), you must:
i) fill in this EULA;
ii) use your official academic email (as data cannot be released to personal emails);
iii) send an email to d.kollias@qmul.ac.uk with subject: C-EXPR-DB request by academic;
iv) include in the email the above signed EULA, the reason why you require access to the database, and your official academic website
If you are a Post-Doc or Ph.D. student: your mentor/advisor/supervisor should perform the steps mentioned above and include you in the list of additional researchers.
If you are from industry and you want to acquire C-EXPR-DB, please email d.kollias@qmul.ac.uk with subject: C-EXPR-DB request from industry and explain the reason why the database access is needed; also specify if it is for research or commercial purposes.
Important notice:
The C-EXPR-DB database is featured in one of the challenges of the recurring ABAW Workshop and Competition, held annually at top-tier conferences. To preserve its integrity, for now, we only share the portion used in the Challenge - 56 videos covering the following seven compound expressions: Fearfully Surprised, Happily Surprised, Sadly Surprised, Disgustedly Surprised, Angrily Surprised, Sadly Fearful, and Sadly Angry. This subset is unannotated and is intended to support the development of methodologies (e.g., supervised, self-supervised, domain adaptation/generalization, zero-/few-shot learning) for recognising these seven compound expressions on a per-frame basis. An evaluation server will be available for testing your developed methods.
We are not responsible for the content nor the meaning of the images within the database.
You agree not to reproduce, duplicate, copy, sell, trade, resell or exploit for any commercial purposes, any portion of the images and any portion of derived data.
You agree not to further copy, publish or distribute any portion of annotations of the dataset. Except, for internal use at a single site within the same organization it is allowed to make copies of the dataset.
We reserve the right to terminate your access to the dataset at any time.
If you use the above data, you must cite the following paper:
D. Kollias: "Multi-Label Compound Expression Recognition: C-EXPR Database & Network". CVPR, 2023
@inproceedings{kollias2023multi, title={Multi-Label Compound Expression Recognition: C-EXPR Database \& Network}, author={Kollias, Dimitrios}, booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition}, pages={5589--5598}, year={2023}}