Personalized situation awareness of drivers (PSAD) dataset is the first public driver behavior dataset integrating situation awareness framework.
(a) four sub-stages of the situation awareness cognitive process: perception, comprehension, projection, and action; (b) scene annotations; (c) driver behavior annotations.
========================================================================================
All contents of PSAD can be downloaded from:
Stimuli videos (Baidu disk): https://pan.baidu.com/s/1MN1-xKBGXleo0AlNYP_5Sg , password: PSAD
For driver attention prediction, including: traffic frames, saliency maps, and fixations maps (Baidu disk): https://pan.baidu.com/s/182EcpwtuJl7XjXl2P9wY6w , password: z2qe
Scene annotations (Google disk): https://drive.google.com/file/d/1fUTxLj-XqQJexKzlz6iJWYRIiw0awlPS/view?usp=sharing
Driver behavior annotations (Google disk): https://drive.google.com/drive/folders/1WDOAExB4NncgAl2-JLIWMlIymtmn-oXB?usp=sharing
======================================================================================
The data source for driver perception-response time on vehicle pre-crash scenario are avaliable at Temporal_annotations_for_PRT.csv
=======================================================================================
Stimuli videos consisted of 2724 real-world accidental videos with 1280×720 resolution and 30 fps, screened from DoTA dataset.
======================================================================================
Scene annotations consisting of anomaly start, collision timestamp, anomaly end, accident type and object_labels for each stimuli video.
- Object_labels comprise object track ID and category, bounding box.
- Note: frame_id in object_labels were labeled at 10 fps, the corrected frame id is equal 3 times original id.
===================================================================================
Driver behavior annotations were collected from six experienced drivers (index: from 100 to 105), which consisting of eye fixations, anomaly category, hazard position, trajectory points, response time, and evasive maneuver.
====================================================================================== Traffic frames, saliency maps, and fixations maps were collected from six experienced drivers (index: from 100 to 105), which consisting of eye fixations, anomaly category, hazard position, trajectory points, response time, and evasive maneuver.
======================================================================================
If you find our dataset is useful, please cite this paper.
@InProceedings{Gan_2021_ITSC,
author = {Shun Gan, Quan Li, Qingfan Wang, WenTao Chen, Detong Qin and Bingbing Nie},
title = {Constructing personalized situation awareness dataset for hazard perception, comprehension, projection, and action of drivers},
booktitle = {The IEEE Conference on Intelligent Transportation Systems,
year = {2021}
}
======================================================================================
WE illustrated the data visualization of some cases from PSAD
p8q77QzOdUs_001954
HNRS3w5zep8_000543
4K_6s1n6BpU_000957