Mimetic Interfaces: Facial Surface EMG Dataset 2015 (c) by Ville Rantanen, Mirja Ilves, Antti Vehkaoja, Anton Kontunen, Jani Lylykangas, Eeva Mäkelä, Markus Rautiainen, Veikko Surakka, and Jukka Lekkala
Mimetic Interfaces: Facial Surface EMG Dataset 2015 is licensed under a Creative Commons Attribution 4.0 International License.
You should have received a copy of the license along with this work. If not, see http://creativecommons.org/licenses/by/4.0/.
The dataset includes Matlab script and functions for processing the data. These are licenced under the MIT license.
The data consists of facial surface EMG signals from the muscles corrugator supercilii, zygomaticus major, orbicularis oris, orbicularis oculi, and masseter.
Fifteen participants (8 females, 7 males) participated in the experiments for data collection. Their age was between 26 and 57 years (mean 40.7, standard deviation 9.6). They all had normal or corrected-to-normal vision and normal hearing by their own report.
The experiments were conducted in two phases. In the A phase, the participant performed voluntary smile, lip pucker, and frown movement tasks while EMG signals from zygomaticus major, orbicularis oris, orbicularis oculi, and corrugator supercilii were measured from the right side of the face. B phase included the measurement of zygomaticus major, orbicularis oris, orbicularis oculi, and masseter from the right side of the face while the participant performed smile and pucker movement tasks while chewing a gum. Both phases had resting tasks between the movement tasks. The resting task was a neutral expression in the first phase, but included chewing in the second one. Phases started with a 1-minute-long resting task. Then 10 repetitions of each movement tasks were performed in randomized order. Each task lasted for 6 seconds. The movements were instructed to be performed as naturally as possible for the time that an on-screen instruction was visible. Instructions regarding movement intensities were not given, and the participants were not informed that their eye blinks will be monitored to avoid causing abnormal blinking.
The used experimental software was E-Prime  stimulation software. The facial surface EMG signals were measured with a NeXus-10 physiological monitoring device by Mind Media BV. The sampling rate was 2048 Hz. Filtering was not set from the measurement software but the hardware is to be expected to have antialiasing filters. The measurements were bipolar using pre-gelled, sintered Ag–AgCl electrodes. A separate grounding electrode was used on the forehead, and the electrodes were placed according to the guidelines of Fridlund and Cacioppo  as shown in the images accompanied with the dataset. Corrugator supercilii was left out from the second phase of the experiments because the measurement device only had 4 channels, and the muscle is not as important in facial pacing as the others are.
The experiments were recorded with a digital video camera at HD quality at 25 fps. The recordings are not published due to privacy issues. They were visually inspected to find out the onset and termination of each eye blink. The beginning each movement task was also determined from the video where the instructions shown to the participant were visible. Screenshots of the videos are included to illustrate the experimental setup. Eye blinks were classified to three categories: ones with a small eyelid movement where the pupil wasn't fully covered, ones where the pupil was fully covered and once where the eye lids was fully closed. Some participants performed multipart blinks where the previous one hadn't ended before the second one started. These are annotated separately in the data.
Schneider, W., Eschman, A., Zuccolotto, A. (2002). E-Prime User’s Guide. Psychology software Tools Inc., Pittsburgh.
A. J. Fridlund and J. T. Cacioppo, “Guidelines for human electromyo- graphic research,” Psychophysiology, vol. 23, no. 5, pp. 567–589, Sep. 1986.
The log files from E-Prime are not published in raw form but they have been read to Matlab and saved as mat-files along with metadata and data from the visual inspection. The log files are named according to the scheme 01A.mat where the number (01-15) is the number of the participant and the letter (A/B) is the experimental phase.
The dataset is accompanied with a Matlab script and functions needed to produce the results for a publication titled "A Survey on the Feasibility of Surface EMG in Facial Pacing" that is to be published at IEEE Engineering in Medicine and Biology Society's EMBC '16 conference.
The included files are:
- Data/*.mat - Matlab data files
- Results/* - Result figures and table as LaTeX tabular
- Screenshots/*.jpg - Images from the experiments
- helper_functions/* - Helper functions for Matlab
- CHANGELOG.txt - Change log to document possible updates
- *_LICENSE.txt - License files for data, metadata, and the Matlab
- README.txt - This document
- Participants.csv - Table with the participants' ages and genders
- dataprocessing.m - The Matlab script for the data processing and outputting the results
- metadata.mat - Metadata with some variables used in the Matlab script