UET Driver Activity Recognition Dataset
This dataset is used in the paper: Driver activity recognition by learning spatiotemporal features of pose and human object interaction
It contains 103 video recordings taken in daylight conditions, comprising 10 actors, mimicking 11 different activities while driving. The length of the videos is variable ranging from 5 sec to 15 sec. Eleven activities are:
- Doze
- Call left
- Call right
- Text right
- Text left
- Safe driving
- Drinking Water
- Reaching hand behind
- Operating radio
- Making hair
- Shifting gear
Sample frames taken from different video clips are shown below:
Please send a request to humza_naveed@yahoo.com for downloading the dataset.
Cite the following article, if you use this dataset.
@article{naveed2021driver,
title={Driver activity recognition by learning spatiotemporal features of pose and human object interaction},
author={Naveed, Humza and Jafri, Fareed and Javed, Kashif and Babri, Haroon Atique},
journal={Journal of Visual Communication and Image Representation},
pages={103135},
year={2021},
publisher={Elsevier}
}