Tracking when the number of individuals in the video frame changes.

<p>(A) Overview of the verification process. The yellow box in the schematic represents the processes using multi-animal tracking tools, while the green box represents the processes using single-animal tracking tools. (B) Schematic diagram of the annotation method. The outlined mouse in the sc...

Full description

Saved in:
Bibliographic Details
Main Author: Hirotsugu Azechi (20700528) (author)
Other Authors: Susumu Takahashi (446938) (author)
Published: 2025
Subjects:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1852022838586769408
author Hirotsugu Azechi (20700528)
author2 Susumu Takahashi (446938)
author2_role author
author_facet Hirotsugu Azechi (20700528)
Susumu Takahashi (446938)
author_role author
dc.creator.none.fl_str_mv Hirotsugu Azechi (20700528)
Susumu Takahashi (446938)
dc.date.none.fl_str_mv 2025-02-10T18:35:25Z
dc.identifier.none.fl_str_mv 10.1371/journal.pbio.3003002.s005
dc.relation.none.fl_str_mv https://figshare.com/articles/figure/Tracking_when_the_number_of_individuals_in_the_video_frame_changes_/28384143
dc.rights.none.fl_str_mv CC BY 4.0
info:eu-repo/semantics/openAccess
dc.subject.none.fl_str_mv Cell Biology
Neuroscience
Immunology
Science Policy
Infectious Diseases
Environmental Sciences not elsewhere classified
Biological Sciences not elsewhere classified
Information Systems not elsewhere classified
findings could enhance
annotation frames needed
sleap &# 8217
unlike physical markers
tracking methods used
virtual markers exist
uses virtual markers
efficiently tackling occlusion
accurate behavioral analysis
animal pose tracking
animal tracking tools
conventional markerless multi
virtual markers
animal tracking
sleap ).
animal videos
animal method
animal deeplabcut
xlink ">
vmtracking ),
social behaviors
simpler yet
labels derived
individual identification
complex naturalistic
attribute features
animals markerless
addressing occlusion
dc.title.none.fl_str_mv Tracking when the number of individuals in the video frame changes.
dc.type.none.fl_str_mv Image
Figure
info:eu-repo/semantics/publishedVersion
image
description <p>(A) Overview of the verification process. The yellow box in the schematic represents the processes using multi-animal tracking tools, while the green box represents the processes using single-animal tracking tools. (B) Schematic diagram of the annotation method. The outlined mouse in the schematic indicates that the mouse was removed from the arena and is absent in the frame. In the case of 3 mice (left), 6 standard annotations were made for each mouse. In the case of 2 mice (center), keypoints for Mouse 2 and Mouse 3 were annotated on each body part of the same mouse. In the case of one mouse (right), keypoints for all mice were annotated on the body parts of that single mouse. Here, the labels are color-coded by individual. (C) Schematic diagram (top) of an experiment that altered the number of mice in the video frame by introducing or removing mice from the arena, and examples of tracking photos during this procedure (bottom). To verify tracking of other mice with leftover keypoints due to the absence of some mice, distances (d: pixels) between identical keypoints for 3 IDs were calculated for each keypoint. The stray keypoints indicated above the outlined mouse, which represents absence in the frame in the top schematic, were evaluated to determine whether they could predict another mouse in the arena, following the annotation method described in (B). Here, the labels are color-coded by body part. (D) If the distance between identical keypoints was within 10 pixels, the 2 ID keypoints were considered overlapping, and the frequency of such frames was calculated for each keypoint and ID pair combination. The diagram illustrates changes across different experimental conditions, with plots indicating the frequency for each keypoint and bars representing each ID pair. The “+” below the diagram indicates the presence of mice in the arena, while “−” indicates their absence. Statistical analysis was conducted using the Friedman test for comparisons within each condition and the Kruskal–Wallis test for comparisons between conditions, with Bonferroni correction applied to account for the 45 total comparison pairs. Additionally, Cliff’s delta was calculated as an effect size for these pairs (<a href="http://www.plosbiology.org/article/info:doi/10.1371/journal.pbio.3003002#pbio.3003002.s013" target="_blank">S1</a> and <a href="http://www.plosbiology.org/article/info:doi/10.1371/journal.pbio.3003002#pbio.3003002.s014" target="_blank">S2</a> Tables). (Ea–c) In Original tracking, markers for Mouse 1 (blue), Mouse 2 (green), and Mouse 3 (red) overlap under various conditions. After removing data from absent mice, a video with no overlaps and accurate tracking by correct IDs is created (delete overlapping keypoints). (Ed) Even in markerless multi-animal pose tracking, markers can overlap. Removing such overlaps allows for creating virtual marker videos with accurate ID identification. In this example, extraneous white data obscuring the correct gray virtual marker are removed. The removal of unnecessary keypoint data is achieved through a Python code that allows specified ranges of tracking data obtained from DeepLabCut to be rewritten as NaN (no data) (<a href="http://www.plosbiology.org/article/info:doi/10.1371/journal.pbio.3003002#pbio.3003002.s019" target="_blank">S1 Protocol</a> and <a href="http://www.plosbiology.org/article/info:doi/10.1371/journal.pbio.3003002#pbio.3003002.s010" target="_blank">S10C Fig</a>). The data underlying these analyses are provided in S1 Data. Additionally, these figures can be reproduced using the MATLAB data and code available in the Zenodo repository (<a href="https://doi.org/10.5281/zenodo.14545410" target="_blank">https://doi.org/10.5281/zenodo.14545410</a>).</p> <p>(TIFF)</p>
eu_rights_str_mv openAccess
id Manara_6cc6bbbc1fbc0610ebfe3edbca3375d2
identifier_str_mv 10.1371/journal.pbio.3003002.s005
network_acronym_str Manara
network_name_str ManaraRepo
oai_identifier_str oai:figshare.com:article/28384143
publishDate 2025
repository.mail.fl_str_mv
repository.name.fl_str_mv
repository_id_str
rights_invalid_str_mv CC BY 4.0
spelling Tracking when the number of individuals in the video frame changes.Hirotsugu Azechi (20700528)Susumu Takahashi (446938)Cell BiologyNeuroscienceImmunologyScience PolicyInfectious DiseasesEnvironmental Sciences not elsewhere classifiedBiological Sciences not elsewhere classifiedInformation Systems not elsewhere classifiedfindings could enhanceannotation frames neededsleap &# 8217unlike physical markerstracking methods usedvirtual markers existuses virtual markersefficiently tackling occlusionaccurate behavioral analysisanimal pose trackinganimal tracking toolsconventional markerless multivirtual markersanimal trackingsleap ).animal videosanimal methodanimal deeplabcutxlink ">vmtracking ),social behaviorssimpler yetlabels derivedindividual identificationcomplex naturalisticattribute featuresanimals markerlessaddressing occlusion<p>(A) Overview of the verification process. The yellow box in the schematic represents the processes using multi-animal tracking tools, while the green box represents the processes using single-animal tracking tools. (B) Schematic diagram of the annotation method. The outlined mouse in the schematic indicates that the mouse was removed from the arena and is absent in the frame. In the case of 3 mice (left), 6 standard annotations were made for each mouse. In the case of 2 mice (center), keypoints for Mouse 2 and Mouse 3 were annotated on each body part of the same mouse. In the case of one mouse (right), keypoints for all mice were annotated on the body parts of that single mouse. Here, the labels are color-coded by individual. (C) Schematic diagram (top) of an experiment that altered the number of mice in the video frame by introducing or removing mice from the arena, and examples of tracking photos during this procedure (bottom). To verify tracking of other mice with leftover keypoints due to the absence of some mice, distances (d: pixels) between identical keypoints for 3 IDs were calculated for each keypoint. The stray keypoints indicated above the outlined mouse, which represents absence in the frame in the top schematic, were evaluated to determine whether they could predict another mouse in the arena, following the annotation method described in (B). Here, the labels are color-coded by body part. (D) If the distance between identical keypoints was within 10 pixels, the 2 ID keypoints were considered overlapping, and the frequency of such frames was calculated for each keypoint and ID pair combination. The diagram illustrates changes across different experimental conditions, with plots indicating the frequency for each keypoint and bars representing each ID pair. The “+” below the diagram indicates the presence of mice in the arena, while “−” indicates their absence. Statistical analysis was conducted using the Friedman test for comparisons within each condition and the Kruskal–Wallis test for comparisons between conditions, with Bonferroni correction applied to account for the 45 total comparison pairs. Additionally, Cliff’s delta was calculated as an effect size for these pairs (<a href="http://www.plosbiology.org/article/info:doi/10.1371/journal.pbio.3003002#pbio.3003002.s013" target="_blank">S1</a> and <a href="http://www.plosbiology.org/article/info:doi/10.1371/journal.pbio.3003002#pbio.3003002.s014" target="_blank">S2</a> Tables). (Ea–c) In Original tracking, markers for Mouse 1 (blue), Mouse 2 (green), and Mouse 3 (red) overlap under various conditions. After removing data from absent mice, a video with no overlaps and accurate tracking by correct IDs is created (delete overlapping keypoints). (Ed) Even in markerless multi-animal pose tracking, markers can overlap. Removing such overlaps allows for creating virtual marker videos with accurate ID identification. In this example, extraneous white data obscuring the correct gray virtual marker are removed. The removal of unnecessary keypoint data is achieved through a Python code that allows specified ranges of tracking data obtained from DeepLabCut to be rewritten as NaN (no data) (<a href="http://www.plosbiology.org/article/info:doi/10.1371/journal.pbio.3003002#pbio.3003002.s019" target="_blank">S1 Protocol</a> and <a href="http://www.plosbiology.org/article/info:doi/10.1371/journal.pbio.3003002#pbio.3003002.s010" target="_blank">S10C Fig</a>). The data underlying these analyses are provided in S1 Data. Additionally, these figures can be reproduced using the MATLAB data and code available in the Zenodo repository (<a href="https://doi.org/10.5281/zenodo.14545410" target="_blank">https://doi.org/10.5281/zenodo.14545410</a>).</p> <p>(TIFF)</p>2025-02-10T18:35:25ZImageFigureinfo:eu-repo/semantics/publishedVersionimage10.1371/journal.pbio.3003002.s005https://figshare.com/articles/figure/Tracking_when_the_number_of_individuals_in_the_video_frame_changes_/28384143CC BY 4.0info:eu-repo/semantics/openAccessoai:figshare.com:article/283841432025-02-10T18:35:25Z
spellingShingle Tracking when the number of individuals in the video frame changes.
Hirotsugu Azechi (20700528)
Cell Biology
Neuroscience
Immunology
Science Policy
Infectious Diseases
Environmental Sciences not elsewhere classified
Biological Sciences not elsewhere classified
Information Systems not elsewhere classified
findings could enhance
annotation frames needed
sleap &# 8217
unlike physical markers
tracking methods used
virtual markers exist
uses virtual markers
efficiently tackling occlusion
accurate behavioral analysis
animal pose tracking
animal tracking tools
conventional markerless multi
virtual markers
animal tracking
sleap ).
animal videos
animal method
animal deeplabcut
xlink ">
vmtracking ),
social behaviors
simpler yet
labels derived
individual identification
complex naturalistic
attribute features
animals markerless
addressing occlusion
status_str publishedVersion
title Tracking when the number of individuals in the video frame changes.
title_full Tracking when the number of individuals in the video frame changes.
title_fullStr Tracking when the number of individuals in the video frame changes.
title_full_unstemmed Tracking when the number of individuals in the video frame changes.
title_short Tracking when the number of individuals in the video frame changes.
title_sort Tracking when the number of individuals in the video frame changes.
topic Cell Biology
Neuroscience
Immunology
Science Policy
Infectious Diseases
Environmental Sciences not elsewhere classified
Biological Sciences not elsewhere classified
Information Systems not elsewhere classified
findings could enhance
annotation frames needed
sleap &# 8217
unlike physical markers
tracking methods used
virtual markers exist
uses virtual markers
efficiently tackling occlusion
accurate behavioral analysis
animal pose tracking
animal tracking tools
conventional markerless multi
virtual markers
animal tracking
sleap ).
animal videos
animal method
animal deeplabcut
xlink ">
vmtracking ),
social behaviors
simpler yet
labels derived
individual identification
complex naturalistic
attribute features
animals markerless
addressing occlusion