Instance segmentation and tracking with cosine embeddings and recurrent hourglass networks

Christian Payer, Darko Štern, Thomas Neff, Horst Bischof, Martin Urschler

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

Abstract

Different to semantic segmentation, instance segmentation assigns unique labels to each individual instance of the same class. In this work, we propose a novel recurrent fully convolutional network architecture for tracking such instance segmentations over time. The network architecture incorporates convolutional gated recurrent units (ConvGRU) into a stacked hourglass network to utilize temporal video information. Furthermore, we train the network with a novel embedding loss based on cosine similarities, such that the network predicts unique embeddings for every instance throughout videos. Afterwards, these embeddings are clustered among subsequent video frames to create the final tracked instance segmentations. We evaluate the recurrent hourglass network by segmenting left ventricles in MR videos of the heart, where it outperforms a network that does not incorporate video information. Furthermore, we show applicability of the cosine embedding loss for segmenting leaf instances on still images of plants. Finally, we evaluate the framework for instance segmentation and tracking on six datasets of the ISBI celltracking challenge, where it shows state-of-the-art performance.

Original languageEnglish
Title of host publicationMedical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings
PublisherSpringer Verlag Heidelberg
Pages3-11
Number of pages9
ISBN (Print)9783030009335
DOIs
Publication statusPublished - 16 Sep 2018
Event21st International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2018 - Granada, Spain
Duration: 16 Sep 201820 Sep 2018

Publication series

NameLecture Notes in Computer Science
Volume11071
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference21st International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2018
CountrySpain
CityGranada
Period16/09/1820/09/18

Fingerprint

Recurrent Networks
Network architecture
Segmentation
Labels
Network Architecture
Semantics
Left Ventricle
Evaluate
Assign
Leaves
Predict
Unit

Keywords

  • Cell
  • Embeddings
  • Instances
  • Recurrent
  • Segmentation
  • Tracking
  • Video

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fields of Expertise

  • Information, Communication & Computing

Cooperations

  • BioTechMed-Graz

Cite this

Payer, C., Štern, D., Neff, T., Bischof, H., & Urschler, M. (2018). Instance segmentation and tracking with cosine embeddings and recurrent hourglass networks. In Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings (pp. 3-11). (Lecture Notes in Computer Science; Vol. 11071 ). Springer Verlag Heidelberg. https://doi.org/10.1007/978-3-030-00934-2_1

Instance segmentation and tracking with cosine embeddings and recurrent hourglass networks. / Payer, Christian; Štern, Darko; Neff, Thomas; Bischof, Horst; Urschler, Martin.

Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings. Springer Verlag Heidelberg, 2018. p. 3-11 (Lecture Notes in Computer Science; Vol. 11071 ).

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

Payer, C, Štern, D, Neff, T, Bischof, H & Urschler, M 2018, Instance segmentation and tracking with cosine embeddings and recurrent hourglass networks. in Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings. Lecture Notes in Computer Science, vol. 11071 , Springer Verlag Heidelberg, pp. 3-11, 21st International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2018, Granada, Spain, 16/09/18. https://doi.org/10.1007/978-3-030-00934-2_1
Payer C, Štern D, Neff T, Bischof H, Urschler M. Instance segmentation and tracking with cosine embeddings and recurrent hourglass networks. In Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings. Springer Verlag Heidelberg. 2018. p. 3-11. (Lecture Notes in Computer Science). https://doi.org/10.1007/978-3-030-00934-2_1
Payer, Christian ; Štern, Darko ; Neff, Thomas ; Bischof, Horst ; Urschler, Martin. / Instance segmentation and tracking with cosine embeddings and recurrent hourglass networks. Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings. Springer Verlag Heidelberg, 2018. pp. 3-11 (Lecture Notes in Computer Science).
@inproceedings{2f299da380de4b8096dc8db4204d9513,
title = "Instance segmentation and tracking with cosine embeddings and recurrent hourglass networks",
abstract = "Different to semantic segmentation, instance segmentation assigns unique labels to each individual instance of the same class. In this work, we propose a novel recurrent fully convolutional network architecture for tracking such instance segmentations over time. The network architecture incorporates convolutional gated recurrent units (ConvGRU) into a stacked hourglass network to utilize temporal video information. Furthermore, we train the network with a novel embedding loss based on cosine similarities, such that the network predicts unique embeddings for every instance throughout videos. Afterwards, these embeddings are clustered among subsequent video frames to create the final tracked instance segmentations. We evaluate the recurrent hourglass network by segmenting left ventricles in MR videos of the heart, where it outperforms a network that does not incorporate video information. Furthermore, we show applicability of the cosine embedding loss for segmenting leaf instances on still images of plants. Finally, we evaluate the framework for instance segmentation and tracking on six datasets of the ISBI celltracking challenge, where it shows state-of-the-art performance.",
keywords = "Cell, Embeddings, Instances, Recurrent, Segmentation, Tracking, Video",
author = "Christian Payer and Darko Štern and Thomas Neff and Horst Bischof and Martin Urschler",
year = "2018",
month = "9",
day = "16",
doi = "10.1007/978-3-030-00934-2_1",
language = "English",
isbn = "9783030009335",
series = "Lecture Notes in Computer Science",
publisher = "Springer Verlag Heidelberg",
pages = "3--11",
booktitle = "Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings",

}

TY - GEN

T1 - Instance segmentation and tracking with cosine embeddings and recurrent hourglass networks

AU - Payer, Christian

AU - Štern, Darko

AU - Neff, Thomas

AU - Bischof, Horst

AU - Urschler, Martin

PY - 2018/9/16

Y1 - 2018/9/16

N2 - Different to semantic segmentation, instance segmentation assigns unique labels to each individual instance of the same class. In this work, we propose a novel recurrent fully convolutional network architecture for tracking such instance segmentations over time. The network architecture incorporates convolutional gated recurrent units (ConvGRU) into a stacked hourglass network to utilize temporal video information. Furthermore, we train the network with a novel embedding loss based on cosine similarities, such that the network predicts unique embeddings for every instance throughout videos. Afterwards, these embeddings are clustered among subsequent video frames to create the final tracked instance segmentations. We evaluate the recurrent hourglass network by segmenting left ventricles in MR videos of the heart, where it outperforms a network that does not incorporate video information. Furthermore, we show applicability of the cosine embedding loss for segmenting leaf instances on still images of plants. Finally, we evaluate the framework for instance segmentation and tracking on six datasets of the ISBI celltracking challenge, where it shows state-of-the-art performance.

AB - Different to semantic segmentation, instance segmentation assigns unique labels to each individual instance of the same class. In this work, we propose a novel recurrent fully convolutional network architecture for tracking such instance segmentations over time. The network architecture incorporates convolutional gated recurrent units (ConvGRU) into a stacked hourglass network to utilize temporal video information. Furthermore, we train the network with a novel embedding loss based on cosine similarities, such that the network predicts unique embeddings for every instance throughout videos. Afterwards, these embeddings are clustered among subsequent video frames to create the final tracked instance segmentations. We evaluate the recurrent hourglass network by segmenting left ventricles in MR videos of the heart, where it outperforms a network that does not incorporate video information. Furthermore, we show applicability of the cosine embedding loss for segmenting leaf instances on still images of plants. Finally, we evaluate the framework for instance segmentation and tracking on six datasets of the ISBI celltracking challenge, where it shows state-of-the-art performance.

KW - Cell

KW - Embeddings

KW - Instances

KW - Recurrent

KW - Segmentation

KW - Tracking

KW - Video

UR - http://www.scopus.com/inward/record.url?scp=85054076332&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-00934-2_1

DO - 10.1007/978-3-030-00934-2_1

M3 - Conference contribution

SN - 9783030009335

T3 - Lecture Notes in Computer Science

SP - 3

EP - 11

BT - Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings

PB - Springer Verlag Heidelberg

ER -