数据集下载地址:
https://ubicomp.eti.uni-siegen.de/home/datasets/icmi18/
数据集对应代码链接:
https://github.com/WJMatthew/WESAD
数据集说明
1.实验设置
参与人员:15人
实验过程:
15名参与人员分别佩戴胸部(chest)和腕部(wrist)传感器,进行90分钟左右的实验,实验阶段主要分为五个:
1.baseline condition: 刚装上传感器后的20分钟,静坐阅读
2.amusement condition:娱乐阶段,观看搞笑短片
3.meditation condition 1:冷静阶段
4.stress condition:压力阶段,实验者处在TSST(Trier Social Stress Test)情境下,这个情景可以有效地让参与者感受到压力
5.meditation condition 2:冷静阶段
ps:每个实验者的实验过程可能有略微不同,具体各阶段的顺序放在各用户目录的SX_quest.csv中,X为用户ID
为了验证实验过程是有效的,每个实验者需要在每个阶段完成后填写调查问卷:
1.Positive and Negative Affect Schedule (PANAS)
PANAS reliably assesses positive (PA) and negative affect (NA),获取二维情绪模型的pleasure值
2.State-Trait Anxiety Inventory (STAI)
gain insight into the anxiety level of the participants
3.Self-Assessment Manikins (SAM)
generate labels in the valence-arousal space,获取二维情绪模型的arousal值
4.Short Stress State Questionnaire (SSSQ)
TSST阶段之后,用SSSQ问卷获取压力类型(worry, engagement, or distress)
具体各问卷问题内容参考数据集readme.pdf => III.2.
2.数据文件说明
# 以S2为例
- S2/
- - S2_E4_Data.zip 存放用户各传感器原始数据
- - - - ACC.csv
- - - - BVP.csv
- - - - EDA.csv
- - - - HR.csv
- - - - IBI.csv
- - - - info.txt
- - - - tags.csv
- - - - TEMP.csv
- - S2.pkl
- - S2_quest.csv 实验阶段和用户问卷数据
- - S2_readme.txt
- - S2_respiban.txt
S2.pkl 存放初步处理过的数据,用python的pickle读取后为字典数据,报错请看[pickle报错解决](https://blog.csdn.net/u012813109/article/details/106966338)
- dict_data
- - subject: "S2"
- - label: 标签 shape=(4255300,) 700hz
- - signal:传感器数据
- - - - chest
- - - - - - ACC:加速度 shape=(4255300,) 700hz
- - - - - - ECG:心电 shape=(4255300,) 700hz
- - - - - - EMG:肌电 shape=(4255300,) 700hz
- - - - - - EDA:皮肤电 shape=(4255300,) 700hz
- - - - - - TEMP:体温 shape=(4255300,) 700hz
- - - - - - RESP:呼吸 shape=(4255300,) 700hz
- - - - wrist
- - - - - - ACC:加速度 shape=(194528,) 32hz
- - - - - - BVP:脉搏 shape=(389056,) 64hz
- - - - - - EDA:皮肤电 shape=(24316,) 4hz
- - - - - - TEMP:体温 shape=(24316,) 4hz
3.标签说明
可以使用情景标签1 2 3 4
也可以使用 用户问卷标签
参考文献
Philip Schmidt, Attila Reiss, Robert Duerichen, Claus Marberger and Kristof Van Laerhoven. 2018. Introducing WESAD, a multimodal dataset for Wearable Stress and Affect Detection. In 2018 International Conference on Multimodal Interaction (ICMI ’18), October 16–20, 2018, Boulder, CO, USA. ACM, New York, NY, USA, 9 pages. https://doi.org/10.1145/3242969.3242985