NITEC: Hand-annotated eye contact data set for interaction from the first-person perspective
Eye contact is a crucial mechanism in non-verbal interaction and plays an important role in our everyday social lives. While humans react very sensitively to eye contact, machines have so far only been able to precisely detect a person's gaze to a limited extent. We take up this challenge and present NITEC, a hand-annotated dataset for eye contact from the first-person perspective. NITEC surpasses existing first-person eye contact datasets in scope and versatility of demographic characteristics, social contexts, and lighting conditions, making it a valuable resource for advancing research in first-person vision-based eye contact. Our extensive evaluations on NITEC show strong performance even on datasets outside the training domain, emphasising its effectiveness and adaptability in different scenarios. This enables seamless application in the fields of computer vision, human-computer interaction and social robotics. We are making our NITEC dataset publicly available to promote reproducibility and support further research in the field of ego-vision interaction.
Translated with DeepL.com (free version)
Publications:
Thorsten Hempel, Magnus Jung, Ahmed A. Abdelrahman, Ayoub Al-Hamadi
NITEC: Versatile Hand-Annotated Eye Contact Dataset for Ego-Vision Interaction,
IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2024
Data Access:
Contact:
Thorsten Hempel, Magnus Jung, Ayoub Al-Hamadi