360_em_algorithms

Here we provide the source code for 5 popular eye movement classification algorithms. These algorithms have been converted in order to work with 360-degree equirectangular gaze recordings.

Moreover we provide a function that reprojects the 360-degree equirectangular data around the equator of the sphere (area with the lowest distortions) and then applies the original algorithms that were developed for monitor based experiments.

1. CONTENT

Before starting using the provided algorithms of this repository you should first clone (or download) the matlab_utils repository that offers utilities for mainly handling ARFF files from here. We also need to clone the matlab_360_utils repository that offers utilities for handling 360-degree data from here. Then add the previous folders to the search path of Matlab with the pathtool or addpath commands.

1.1 Converted Algorithms

The algorithms that have word File at the end take a file as an input and store the result as a new attributes at the output file. The algorithms with the same basename work on preloaded data. The algorithms that require few parameters, they are provided directly as input arguments. For the more complex algorithms the parameters are stored in a file and are loaded from it. For all the algorithms we provide default parameters which were provided by the original authors except for one parameter in the Larsson et al. (2015) algorithm which is explained in our paper.

Another distinct functionality of the eye movement classification algorithms that are provided here is their ability to distinguish between eye together gaze motion (E+H) eye in head gaze motion (field-of-view, FOV). This can be set in the typeOfMotion parameter.

The 360-degree ware implementation of the Larsson et al. (2015) is provided in its own directory and the algorithm names follow the same convention.

The list of available algorithms is given below

File	Use
DetectSaccades360.m	runs the saccade detector from Dorr et al. (2010) on loaded data and returns a logical vector with true where a saccade was detected
DetectSaccades360IVT.m	runs the I-VT saccade detector from Salvucci and Goldberg (2000) on loaded data and returns a logical vector with true where a saccade was detected
DetectFixations360.m	runs the fixation detector from Dorr et al. (2010) on loaded data and returns a logical vector with true where a fixation was detected
DetectSaccades360IDT.m	runs the I-DT saccade detector from Salvucci and Goldberg (2000) on loaded data and returns a logical vector with true where a fixation was detected
DetectBlinks360.m	detects blinks based on noise intervals and the distance of saccade intervals from them and returns a logical vector with true where a blink was detected
larsson360/DetectLarsson360.m	runs the Larsson et al (2015)detection on loaded data and returns a column vector with the detected eye movements
DetectSaccades360File.m	runs the saccade detector from Dorr et al. (2010) for one file and stores the result in another one
DetectSaccades360IVTFile.m	runs the I-VT saccade detector from Salvucci and Goldberg (2000) for one file and stores the result in another one
DetectFixations360File.m	runs the fixation detector from Dorr et al. (2010) for one file and stores the result in another one
DetectSaccades360IDTFile.m	runs the I-DT saccade detector from Salvucci and Goldberg (2000) for one file and stores the result in another one
DetectBlinks360File.m	runs the above blink detector for one file and stores the result in another one
larsson_360/DetectLarsson360File.m	runs the Larsson et al. (2015) detection for one file and stores the result in another one
larsson360/DetectLarsson360FileRegex.m	runs the Larsson detection for all files matched by the wildcard-regex (ex. '../GazeCom/gaze_arff//.arff')
params_saccades_equirect.txt	file containing the parameters that are used for saccade and blink detection per Dorr et al. (2010). The speed thresholds where for the eye+head representation
params_saccades_fov.txt	file containing the parameters that are used for saccade and blink detection er Dorr et al. (2010). The speed thresholds were optimized for eye within head representation
params_fixation.txt	file containing the parameters that are used for fixation detection per Dorr et al. (2010)
larsson_360/params_larsson.txt	file containing the parameters that are used for SP and fixation detection per Larsson et al. (2015)

1.2 Data Reprojection

If the conversion of an algorithm is not easy because it is either very complex or it will be applied to small amount of data, we can reproject the equirectangular data to areas with low distortions and apply the original algorithms directly. Here we also offer the possibility of distinguishing between E+H and FOV gaze motion as in the case of the converted algorithms with the DataReprojDetection.m and DataReprojFovDetection.m functions.

The main idea behind reprojection is to provide the eye movement detection implementation function name as input to the reprojection detection functions and then call them with converted data as input. The detection functions should take as input at least the data, metadata, and attributes as returned from LoadArff.m. All the extra arguments can provided through the varargin input argument. Their output comprises of a column vector with integer when an eye movement is detected.

A more detailed explanation of the input arguments is given below

Input arguments	Use
arffFile	ARFF file to process
outFile	file name to store result
outputAtt	name of the attribute in the output ARFF that holds detected eye movements
attValues	nominal values of the added attributes as returned from eye movement detection algorithm. They are a string in the form '{unassigned, fixation}' if the detection algorithm returns 0 for unassigned and 1 for fixations
detFuncName	detection function name as string. Ex. 'DetectSaccadesIVVT'
varargin	required extra arguments for calling the detection function. The data, metadata, attributes are passed to the detection function by default in this order followed by the varargin arguments

The list of used files for data reprojection is given below

File	Use
DataReprojDetection.m	(main function) calls the provided eye movement detection function on E+H (eye and head) motion data
DataReprojFovDetection.m	(main function) calls the provided eye movement detection function on FOV (eye within head) motion data
ProjectEquirect2Fov.m	projects data to the field-of-view
ProjectEquirect2FovFile.m	projects data to the field-of-view and stores it to a file

2. DATA FORMAT

All the function use the ARFF data format for input and output to the disk. The initial ARFF format was extended as described in Agtzidis et al. (2016) and was further expanded for 360-degree gaze data.

Here the "@relation" is set to gaze_360 to distinguish the recordings from plain gaze recordings. We also make use of the "%@METADATA" special comments which describe the field of view of the used headset. Apart from the default metadata width_px, height_px, distance_mm, width_mm, height_mm we also use the extra metadata fov_width_px, fov_width_deg, fov_height_px, fov_height_deg that describe the headset properties.

The "@attribute" x and y represent the gaze coordinates in the equirectangular space (i.e. the head and eye position together). The x_head and y_head attributes represent head coordinates in the equirectangular space. The angle_deg_head attribute is equivalent to the roll principal axis.

2.1 ARFF Example

@RELATION gaze_360

%@METADATA distance_mm 0.00
%@METADATA height_mm 0.00
%@METADATA height_px 1080
%@METADATA width_mm 0.00
%@METADATA width_px 1920

%@METADATA fov_height_deg 100.00
%@METADATA fov_height_px 1440
%@METADATA fov_width_deg 100.00
%@METADATA fov_width_px 1280

@ATTRIBUTE time INTEGER
@ATTRIBUTE x NUMERIC
@ATTRIBUTE y NUMERIC
@ATTRIBUTE confidence NUMERIC
@ATTRIBUTE x_head NUMERIC
@ATTRIBUTE y_head NUMERIC
@ATTRIBUTE angle_deg_head NUMERIC
@ATTRIBUTE labeller_1 {unassigned,fixation,saccade,SP,noise,VOR,OKN}


@DATA
0,960.00,540.00,1.00,960.00,540.00,1.22,fixation
5000,959.00,539.00,1.00,959.00,539.00,1.23,fixation
13000,959.00,539.00,1.00,959.00,539.00,1.23,fixation
18000,959.00,539.00,1.00,959.00,539.00,1.23,fixation
29000,959.00,539.00,1.00,959.00,539.00,1.24,fixation
34000,959.00,539.00,1.00,959.00,539.00,1.24,fixation
45000,959.00,539.00,1.00,959.00,539.00,1.24,fixation
49000,959.00,539.00,1.00,959.00,539.00,1.24,fixation
61000,959.00,539.00,1.00,959.00,539.00,1.24,fixation
66000,959.00,539.00,1.00,959.00,539.00,1.24,fixation
77000,959.00,539.00,1.00,959.00,540.00,1.24,fixation
82000,959.00,539.00,1.00,959.00,540.00,1.24,fixation
94000,959.00,539.00,1.00,960.00,540.00,1.24,fixation
99000,959.00,539.00,1.00,960.00,540.00,1.24,fixation
110000,959.00,539.00,1.00,960.00,540.00,1.25,fixation
114000,959.00,539.00,1.00,960.00,540.00,1.25,fixation
125000,958.00,538.00,1.00,960.00,540.00,1.26,saccade
129000,956.00,537.00,1.00,960.00,540.00,1.27,saccade
141000,948.00,530.00,1.00,960.00,540.00,1.28,saccade

2.2 Recovery of HMD Pose

The x_head, y_head, angle_deg_head allow us to recover the headset pose because we use 360-degree equirectangular videos and therefore exist no translations during video presentation. For understanding the process take a look at functions HeadToVideoRot.m and YZXrotation.m in the matlab_360_utils repository.

3. GENERAL INFORMATION

Author: Ioannis Agtzidis Contact: ioannis.agtzidis@tum.de

If you use any algorithm re-implementation for 360-degree equirectangular stimuli, please cite:

\@inproceedings{agtzidis2019conversion,
author = {Agtzidis, Ioannis and Dorr, Michael},
title = {Getting (More) Real: Bringing Eye Movement Classification to HMD Experiments with Equirectangular Stimuli},
booktitle = {Proceedings of the 11th ACM Symposium on Eye Tracking Research & Applications},
series = {ETRA '19},
year = {2019},
isbn = {978-1-4503-6709-7},
location = {Denver, Colorado},
pages = {18:1--18:8},
articleno = {18},
numpages = {8},
url = {http://doi.acm.org/10.1145/3314111.3319829},
doi = {10.1145/3314111.3319829},
acmid = {3319829},
publisher = {ACM},
address = {New York, NY, USA},
keywords = {360{\${}^\circ\$} content, event detection, eye movement classification},
}

4. REFERENCES

Ioannis Agtzidis, Mikhail Startsev, and Michael Dorr. 2016. In the pursuit of (ground) truth: A hand-labelling tool for eye movements recorded during dynamic scene viewing. In 2016 IEEE Second Workshop on Eye Tracking and Visualization (ETVIS). 65–68. https://doi.org/10.1109/ETVIS.2016.7851169

Michael Dorr, Thomas Martinetz, Karl R Gegenfurtner, and Erhardt Barth. 2010. Variability of eye movements when viewing dynamic natural scenes. Journal of Vision 10, 10 (2010), 28–28

Linnéa Larsson, Marcus Nyström, Richard Andersson, and Martin Stridh. 2015. Detection of fixations and smooth pursuit movements in high-speed eye-tracking data. Biomedical signal processing and control 18 (2015), 145–152.