AVDOS-VR: Affective Video Database for VR using subjective ratings and physiological data recorded with wearable EmteqPro system
- Citation Author(s):
- Submitted by:
- Michal Gnacek
- Last updated:
- Thu, 03/30/2023 - 13:53
- Data Format:
Physiological database including 7 pairs of facial surface EMG (left/right frontalis, left/right zygomaticus, left/right orbicularis and centre corrugator) muscles, forehead PPG, IMU (Accelerometer, Gyroscope, Magnetometer) and continuous self-ratings of arousal and valence while watching video stimuli. 37 participants watched an average of 23 minutes worth of videos. This included blocks of positive, negative and neutral segments. Each segment included 10 videos, 30 seconds long each. Relaxation video was played between each segment. Order of segments and videos within each segment was random.
AVDOS-VR uses a subset of videos published in peer-reviewed AVDOS database (https://ieeexplore.ieee.org/document/9953891).
"Dataset_AVDOSVR_postprocessed.csv" - is a processed data file as an optional download for those wishing to get straight into analysis. This file will be automatically generated as part of the notebook #1
"events.table.pdf" - contains a table listing all types of possible events recordoded as part of the data collection method and saved in the events json file
"video_ratings.pdf" - contains a table with mean valence and arousal ratings for each video as well as number of ratings provided by all participants
Data folder for each participants consists of ".csv", ".json" and ".raw" files.
Json files contain event data (for example start/end of each video) that can be synchronised with physiological data.
Raw files contain proprietary data format for physiological data saved using emteqPro device. These files can be converted into csv files using DabToCsv converted. Read more at: https://support.emteqlabs.com/downloads/
CSV files contain converted raw files (physiological data, EMG, PPG, IMU)
- Clone code repository from: https://github.com/michalgnacek/AVDOS-VR
- Download the data and paste entire data folder into root folder generating the tree AVDOS-VR/data/participant_XXX/*
- To see example analysis, check the files in AVDOS-VR/notebooks/*
- If using root folder as the working directory, exclude the following file types from IDE search function to avoid searching through large files (*.csv, *.raw, *.json)
Data can also be downloaded from gnacek.com/affective-video-database-online-study
Processed data is available at the link above. It is an optional download as this file will be generated automatically when running preprocessing notebook 1
- github link Python Library Git Repository.txt (40 bytes)
- tables for video ratings and event types ratings and event types.zip (109.90 kB)
- Dataset_AVDOSVR_postprocessed.csv (772.93 MB)
- data.zip (3.06 GB)