Real World Engagement Attention Dataset
(RWEAD)
Last updated 2026/10/12
RWEAD #2 is a multimodal indoor-behavior dataset acquired in a 7.8 m × 6.6 m room configured as an Indoor-category exhibition stand within a building hall. The sensing infrastructure comprises 11 devices spanning five modalities: six RGB-D cameras, one thermal camera, two millimeter-wave (mmWave) sensors, one LiDAR, and one microphone. Three points of interest (PoIs)—Table, Model, and TV—are defined within the room. The participant cohort is male-only and consists of amateur actors aged 20–30, all university students without formal acting training. Each participant received a script and was allotted five minutes for memorization prior to data acquisition. The dataset covers the following behavioral, affective, spatial, and interactional constructs and key performance indicators (KPIs): engagement, sentiment, density maps, emotion, motion patterns, attention, person counting, dwell times, conversion, conversation, retention, feedback, and social interaction.
Layout of the room (AI rendering)
Acquisition Display and Environment Photos
Dataset Protocol
Constructs and KPIs
RWEAD #2 download available at OSF - https://doi.org/10.17605/OSF.IO/EJQSA
For RWEAD #2 cite (temporary):
Rodrigues, J. M. F., Lemos, M., El Afia, M., Solá, D., Walid, A., Santos, R., António, G., Cardoso, P. J. S., & Martins, J. A. (2026). MArtAE: A unified architecture for real-time audience experience analytics at live events. Accepted to the 8th International Conference on Soft Computing and its Engineering Applications (icSoftComp2026), Singapore.
ALGARVE 2030, Portugal 2030 and by the European Union, ALGARVE-FEDER-02964500, Ref. 24298 - Project AI.INSIGHTSI