CogBeacon
收藏资源简介：
CogBeacon是一个多模态数据集，旨在针对人类性能中的认知疲劳效应。该数据集包含从19名男女用户收集的76个会话，这些用户执行了不同版本的威斯康星卡片分类测试（WCST）。在每个会话中，我们记录并完全注释了用户的EEG功能、面部关键点、实时自我报告的认知疲劳以及认知任务期间实现的表现指标（成功率、响应时间、错误数量等）的详细信息。此外，我们还提供了一个基线机器学习分析，用于预测认知疲劳，以及我们的多模态WCST实现，以允许其他研究者扩展或修改CogBeacon数据收集框架的功能。据我们所知，这是第一个专门设计来评估认知疲劳的多模态数据集。
CogBeacon is a multimodal dataset specifically designed to address the effects of cognitive fatigue in human performance. The dataset comprises 76 sessions collected from 19 male and female users who performed various versions of the Wisconsin Card Sorting Test (WCST). During each session, we recorded and fully annotated the users' EEG features, facial key points, real-time self-reported cognitive fatigue, and detailed performance metrics (such as success rate, response time, and number of errors) achieved during cognitive tasks. Additionally, we provide a baseline machine learning analysis for predicting cognitive fatigue, along with our multimodal WCST implementation, to allow other researchers to extend or modify the functionality of the CogBeacon data collection framework. To the best of our knowledge, this is the first multimodal dataset specifically designed to evaluate cognitive fatigue.
CogBeacon: A Multi-Modal Dataset for Modeling Cognitive Fatigue
Overview
CogBeacon is a multi-modal dataset designed to assess cognitive fatigue in human performance. It includes 76 sessions from 19 users performing the Wisconsin Card Sorting Test (WCST), a cognitive test that evaluates cognitive flexibility and reasoning. Data collected during these sessions include EEG functionality, facial keypoints, real-time self-reports on cognitive fatigue, and performance metrics.
Dataset Structure
The dataset is organized into four main folders:
-
EEG Data:
- Filename Structure: Each session is stored in a separate folder named "user_<userID> _ <StimuliType> _ <GameMode>".
- Data Types: Raw EEG, Absolute Frequency Bands, Relative Frequency Bands, Session Score for each Frequency band, Signal Quality Indicator.
- Equipment: Data recorded using the Muse EEG headset.
-
Facial Keypoints:
- Filename Structure: Data stored in the "face_keypoints" folder, with filenames structured as <round_under_the_same_rule> _ <roundID> _ <frameID>.
- Data Collection: Captured using a webcam at 2 FPS, employing a Regression Tree approach for keypoint identification.
-
Fatigue Self Report:
- Filename Structure: Data stored in the "fatigue_self_report" folder as CSV files, with filenames following the same structure as other data types.
- Data Content: Records the total number of times a user pressed a button to indicate cognitive fatigue.
-
User Performance:
- Filename Structure: Data stored as CSV files in the "user_performance" folder, with filenames structured similarly to other data types.
- Metrics Included: Round Number, Question Number, Level, Score, Stimuli, Stimuli Type, Response, Time, Correct, NON-PER Errors, PER Errors.
Additional Resources
- Machine Learning Analysis: Code and data used for the ML analysis can be found HERE.
- EEG Data Codes: Python codes for EEG data processing are available HERE.
Confidentiality & Data Sharing
The dataset was approved by the Institutional Review Board (IRB) of the University of Texas at Arlington. For inquiries regarding confidentiality or data sharing, contact the IRB office at UTA or the project personnel.
