Affective Software Engineering research group
Two datasets of spontaneous facial expressions from students solving programming tasks — DevEmo (2023) and DevEmo+ (2026) — together covering 521 video clips from 84 participants.
DevEmo (2023) and DevEmo+ (2026) are companion datasets of students' facial expressions while solving programming tasks. DevEmo contributed 217 clips from 33 participants; DevEmo+ extended the work with recordings collected "in the wild," from participants' own computing environments, adding 304 clips from 51 participants, balanced between 152 emotional and 152 neutral expressions. Together they cover 521 clips from 84 participants.
Together, DevEmo and DevEmo+ can be used as a training and test set for models that recognise the emotions of computer users, and to study how confusion, frustration and other affective states unfold as programmers work. They are well suited to research on personalised, affect-aware tools for computer-based learning environments.
Both datasets were annotated with the help of AffecTube, a crowdsourcing tool built by the same research group for video-based emotion annotation.
Get the data
DevEmo and DevEmo+ are distributed separately, each in the way that suits its licence — together they give you the full 521-clip, 84-participant corpus.
2026 · Open access
304 clips · 51 participants · openly available, licensed under CC BY-NC 4.0 (attribution required, non-commercial use only).
2023 · By request
217 clips · 33 participants · available on request. Send an email with the subject "DevEmo dataset access request" to michal.wrobel@pg.edu.pl.
The dataset is intended for the scientific community — please send requests from an academic institution address, or describe in the message body what the dataset will be used for.
DevEmo and DevEmo+ were developed together as part of research conducted by the Affective Software Engineering group.
What the recordings look like
Cite this work
DevEmo and DevEmo+ are described in two papers — cite whichever matches the data you use, or both if you draw on the combined corpus.
DOI 10.1038/s41597-026-08065-7
@article{wrobel2026devemoplus,
author = {Wróbel, MR and Barkana, DE and Leszczyński, S},
title = {{DevEmo+: A Video Dataset of Spontaneous Facial Expressions of Students Solving Programming Tasks in the Wild}},
journal = {Scientific Data},
year = {2026},
doi = {10.1038/s41597-026-08065-7},
url = {https://doi.org/10.1038/s41597-026-08065-7}
}DevEmo — Software Developers' Facial Expression Dataset
@article{manikowska2023devemo,
author = {Manikowska, M and Sadowski, D and Sowinski, A and Wróbel, MR},
title = {{DevEmo — Software Developers' Facial Expression Dataset}},
journal = {Applied Sciences},
year = {2023},
doi = {10.3390/app13063839},
url = {https://www.mdpi.com/2076-3417/13/6/3839}
}