Adversarial learning for semi-supervised semantic segmentation

Wei Chih Hung, Yi Hsuan Tsai, Yan Ting Liou, Yen Yu Lin, Ming Hsuan Yang

Research output: Contribution to conferencePaper

4 Citations (Scopus)

Abstract

We propose a method for semi-supervised semantic segmentation using an adversarial network. While most existing discriminators are trained to classify input images as real or fake on the image level, we design a discriminator in a fully convolutional manner to differentiate the predicted probability maps from the ground truth segmentation distribution with the consideration of the spatial resolution. We show that the proposed discriminator can be used to improve semantic segmentation accuracy by coupling the adversarial loss with the standard cross entropy loss of the proposed model. In addition, the fully convolutional discriminator enables semi-supervised learning through discovering the trustworthy regions in predicted results of unlabeled images, thereby providing additional supervisory signals. In contrast to existing methods that utilize weakly-labeled images, our method leverages unlabeled images to enhance the segmentation model. Experimental results on the PASCAL VOC 2012 and Cityscapes datasets demonstrate the effectiveness of the proposed algorithm.

Original languageEnglish
Publication statusPublished - 2019 Jan 1
Event29th British Machine Vision Conference, BMVC 2018 - Newcastle, United Kingdom
Duration: 2018 Sep 32018 Sep 6

Conference

Conference29th British Machine Vision Conference, BMVC 2018
CountryUnited Kingdom
CityNewcastle
Period18/9/318/9/6

Fingerprint

Discriminators
Semantics
Supervised learning
Volatile organic compounds
Entropy

All Science Journal Classification (ASJC) codes

  • Computer Vision and Pattern Recognition

Cite this

Hung, W. C., Tsai, Y. H., Liou, Y. T., Lin, Y. Y., & Yang, M. H. (2019). Adversarial learning for semi-supervised semantic segmentation. Paper presented at 29th British Machine Vision Conference, BMVC 2018, Newcastle, United Kingdom.
Hung, Wei Chih ; Tsai, Yi Hsuan ; Liou, Yan Ting ; Lin, Yen Yu ; Yang, Ming Hsuan. / Adversarial learning for semi-supervised semantic segmentation. Paper presented at 29th British Machine Vision Conference, BMVC 2018, Newcastle, United Kingdom.
@conference{2a161159cea542cf92ec93dbdf660b47,
title = "Adversarial learning for semi-supervised semantic segmentation",
abstract = "We propose a method for semi-supervised semantic segmentation using an adversarial network. While most existing discriminators are trained to classify input images as real or fake on the image level, we design a discriminator in a fully convolutional manner to differentiate the predicted probability maps from the ground truth segmentation distribution with the consideration of the spatial resolution. We show that the proposed discriminator can be used to improve semantic segmentation accuracy by coupling the adversarial loss with the standard cross entropy loss of the proposed model. In addition, the fully convolutional discriminator enables semi-supervised learning through discovering the trustworthy regions in predicted results of unlabeled images, thereby providing additional supervisory signals. In contrast to existing methods that utilize weakly-labeled images, our method leverages unlabeled images to enhance the segmentation model. Experimental results on the PASCAL VOC 2012 and Cityscapes datasets demonstrate the effectiveness of the proposed algorithm.",
author = "Hung, {Wei Chih} and Tsai, {Yi Hsuan} and Liou, {Yan Ting} and Lin, {Yen Yu} and Yang, {Ming Hsuan}",
year = "2019",
month = "1",
day = "1",
language = "English",
note = "29th British Machine Vision Conference, BMVC 2018 ; Conference date: 03-09-2018 Through 06-09-2018",

}

Hung, WC, Tsai, YH, Liou, YT, Lin, YY & Yang, MH 2019, 'Adversarial learning for semi-supervised semantic segmentation', Paper presented at 29th British Machine Vision Conference, BMVC 2018, Newcastle, United Kingdom, 18/9/3 - 18/9/6.

Adversarial learning for semi-supervised semantic segmentation. / Hung, Wei Chih; Tsai, Yi Hsuan; Liou, Yan Ting; Lin, Yen Yu; Yang, Ming Hsuan.

2019. Paper presented at 29th British Machine Vision Conference, BMVC 2018, Newcastle, United Kingdom.

Research output: Contribution to conferencePaper

TY - CONF

T1 - Adversarial learning for semi-supervised semantic segmentation

AU - Hung, Wei Chih

AU - Tsai, Yi Hsuan

AU - Liou, Yan Ting

AU - Lin, Yen Yu

AU - Yang, Ming Hsuan

PY - 2019/1/1

Y1 - 2019/1/1

N2 - We propose a method for semi-supervised semantic segmentation using an adversarial network. While most existing discriminators are trained to classify input images as real or fake on the image level, we design a discriminator in a fully convolutional manner to differentiate the predicted probability maps from the ground truth segmentation distribution with the consideration of the spatial resolution. We show that the proposed discriminator can be used to improve semantic segmentation accuracy by coupling the adversarial loss with the standard cross entropy loss of the proposed model. In addition, the fully convolutional discriminator enables semi-supervised learning through discovering the trustworthy regions in predicted results of unlabeled images, thereby providing additional supervisory signals. In contrast to existing methods that utilize weakly-labeled images, our method leverages unlabeled images to enhance the segmentation model. Experimental results on the PASCAL VOC 2012 and Cityscapes datasets demonstrate the effectiveness of the proposed algorithm.

AB - We propose a method for semi-supervised semantic segmentation using an adversarial network. While most existing discriminators are trained to classify input images as real or fake on the image level, we design a discriminator in a fully convolutional manner to differentiate the predicted probability maps from the ground truth segmentation distribution with the consideration of the spatial resolution. We show that the proposed discriminator can be used to improve semantic segmentation accuracy by coupling the adversarial loss with the standard cross entropy loss of the proposed model. In addition, the fully convolutional discriminator enables semi-supervised learning through discovering the trustworthy regions in predicted results of unlabeled images, thereby providing additional supervisory signals. In contrast to existing methods that utilize weakly-labeled images, our method leverages unlabeled images to enhance the segmentation model. Experimental results on the PASCAL VOC 2012 and Cityscapes datasets demonstrate the effectiveness of the proposed algorithm.

UR - http://www.scopus.com/inward/record.url?scp=85069792255&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85069792255&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:85069792255

ER -

Hung WC, Tsai YH, Liou YT, Lin YY, Yang MH. Adversarial learning for semi-supervised semantic segmentation. 2019. Paper presented at 29th British Machine Vision Conference, BMVC 2018, Newcastle, United Kingdom.