Two hours in Hollywood: A manually annotated ground truth data set of eye movements during movie clip watching

In this short article we present our manual annotation of the eye movement events in a subset of the large-scale eye tracking data set Hollywood2. Our labels include fixations, saccades, and smooth pursuits, as well as a noise event type (the latter representing either blinks, loss of tracking, or p...

Full description

Saved in:
Bibliographic Details
Main Authors: Ioannis Agtzidis, Mikhail Startsev, Michael Dorr
Format: Article
Language:English
Published: MDPI AG 2020-12-01
Series:Journal of Eye Movement Research
Subjects:
Online Access:https://bop.unibe.ch/JEMR/article/view/6008
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850189722933002240
author Ioannis Agtzidis
Mikhail Startsev
Michael Dorr
author_facet Ioannis Agtzidis
Mikhail Startsev
Michael Dorr
author_sort Ioannis Agtzidis
collection DOAJ
description In this short article we present our manual annotation of the eye movement events in a subset of the large-scale eye tracking data set Hollywood2. Our labels include fixations, saccades, and smooth pursuits, as well as a noise event type (the latter representing either blinks, loss of tracking, or physically implausible signals). In order to achieve more consistent annotations, the gaze samples were labelled by a novice rater based on rudimentary algorithmic suggestions, and subsequently corrected by an expert rater. Overall, we annotated eye movement events in the recordings corresponding to 50 randomly selected test set clips and 6 training set clips from Hollywood2, which were viewed by 16 observers and amount to a total of approximately 130 minutes of gaze data. In these labels, 62.4% of the samples were attributed to fixations, 9.1% – to saccades, and, notably, 24.2% – to pursuit (the remainder marked as noise). After evaluation of 15 published eye movement classification algorithms on our newly collected annotated data set, we found that the most recent algorithms perform very well on average, and even reach human-level labelling quality for fixations and saccades, but all have a much larger room for improvement when it comes to smooth pursuit classification. The data set is made available at https://gin.g- node.org/ioannis.agtzidis/hollywood2_em.
format Article
id doaj-art-2e0a95f16e4b4b9eb79641427e5d8710
institution OA Journals
issn 1995-8692
language English
publishDate 2020-12-01
publisher MDPI AG
record_format Article
series Journal of Eye Movement Research
spelling doaj-art-2e0a95f16e4b4b9eb79641427e5d87102025-08-20T02:15:32ZengMDPI AGJournal of Eye Movement Research1995-86922020-12-0113410.16910/jemr.13.4.5Two hours in Hollywood: A manually annotated ground truth data set of eye movements during movie clip watchingIoannis Agtzidis0Mikhail Startsev1Michael Dorr2Technical University of Munich, GermanyTechnical University of Munich, GermanyTechnical University of Munich, GermanyIn this short article we present our manual annotation of the eye movement events in a subset of the large-scale eye tracking data set Hollywood2. Our labels include fixations, saccades, and smooth pursuits, as well as a noise event type (the latter representing either blinks, loss of tracking, or physically implausible signals). In order to achieve more consistent annotations, the gaze samples were labelled by a novice rater based on rudimentary algorithmic suggestions, and subsequently corrected by an expert rater. Overall, we annotated eye movement events in the recordings corresponding to 50 randomly selected test set clips and 6 training set clips from Hollywood2, which were viewed by 16 observers and amount to a total of approximately 130 minutes of gaze data. In these labels, 62.4% of the samples were attributed to fixations, 9.1% – to saccades, and, notably, 24.2% – to pursuit (the remainder marked as noise). After evaluation of 15 published eye movement classification algorithms on our newly collected annotated data set, we found that the most recent algorithms perform very well on average, and even reach human-level labelling quality for fixations and saccades, but all have a much larger room for improvement when it comes to smooth pursuit classification. The data set is made available at https://gin.g- node.org/ioannis.agtzidis/hollywood2_em.https://bop.unibe.ch/JEMR/article/view/6008Eye trackingeye movementgazesmooth pursuiteye movement classificationhand-labelling
spellingShingle Ioannis Agtzidis
Mikhail Startsev
Michael Dorr
Two hours in Hollywood: A manually annotated ground truth data set of eye movements during movie clip watching
Journal of Eye Movement Research
Eye tracking
eye movement
gaze
smooth pursuit
eye movement classification
hand-labelling
title Two hours in Hollywood: A manually annotated ground truth data set of eye movements during movie clip watching
title_full Two hours in Hollywood: A manually annotated ground truth data set of eye movements during movie clip watching
title_fullStr Two hours in Hollywood: A manually annotated ground truth data set of eye movements during movie clip watching
title_full_unstemmed Two hours in Hollywood: A manually annotated ground truth data set of eye movements during movie clip watching
title_short Two hours in Hollywood: A manually annotated ground truth data set of eye movements during movie clip watching
title_sort two hours in hollywood a manually annotated ground truth data set of eye movements during movie clip watching
topic Eye tracking
eye movement
gaze
smooth pursuit
eye movement classification
hand-labelling
url https://bop.unibe.ch/JEMR/article/view/6008
work_keys_str_mv AT ioannisagtzidis twohoursinhollywoodamanuallyannotatedgroundtruthdatasetofeyemovementsduringmovieclipwatching
AT mikhailstartsev twohoursinhollywoodamanuallyannotatedgroundtruthdatasetofeyemovementsduringmovieclipwatching
AT michaeldorr twohoursinhollywoodamanuallyannotatedgroundtruthdatasetofeyemovementsduringmovieclipwatching