Multimodal Datasets for Driver Monitoring

Comprehensive multimodal data set for mental overload (CMMO)

Fraunhofer IIS together with University Erlangen-Nuremberg is creating unique gold-standard data set enabling creation of ML models (»CMMO«).

Combining competencies from psychology, biophysics and computer science, we create an extensive database of psychophysiological data on the state of mental overload.

Comprehensive subject-based study with different stimuli and modalities to assess psychophysiological conditions
Representative data with a broad range of test persons
Scientifically proven data quality with validated ground truth
Synchronized data base of various modalities

The Challenges

A comprehensive database of high quality scientific and multimodal data is essential for the interpretation of human behavior with AI.
However, existing data sets for measuring cognitive load and other affective states have severe drawbacks:

Imbalanced populations
Inadequate stimuli and study setup
Invalid ground truth
Low quality measurements
Missing/inaccurate synch between modalities
and/or stimuli

Modalities and Stimuli for Psycho-physiological Datasets

Various stimuli to trigger mental overload	EEG: setting ground truth for mental overload
Muscle activity with EMG	Electrodermal response (EDR)
Breathing	Photoplethysmography
Skin temperature	Electrogastrogram
HR, HRV using ECG	Camera based HR analysis (RGB, IR)
Camera based analysis of facial expressions	Eyetracking

CMMO will be available this year and enable building Machine Learning models predicting mental (over)load based on selected modalities.

Call us for more information and license conditions!

Multimodal Data Sets for Driver Monitoring