TY - GEN
T1 - Alleviating semantic-level shift
T2 - 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, CVPRW 2020
AU - Wang, Zhonghao
AU - Wei, Yunchao
AU - Feris, Rogerio
AU - Xiong, Jinjun
AU - Hwu, Wen Mei
AU - Huang, Thomas S
AU - Shi, Honghui
N1 - Funding Information:
Acknowledgment This work is supported by IBM-UIUC Center for Cognitive Computing Systems Research(C3SR).
Publisher Copyright:
© 2020 IEEE.
PY - 2020/6
Y1 - 2020/6
N2 - Utilizing synthetic data for semantic segmentation can significantly relieve human efforts in labelling pixel-level masks. A key challenge of this task is how to alleviate the data distribution discrepancy between the source and target domains, i.e. reducing domain shift. The common approach to this problem is to minimize the discrepancy between feature distributions from different domains through adversarial training. However, directly aligning the feature distribution globally cannot guarantee consistency from a local view (i.e. semantic-level). To tackle this issue, we propose a semi-supervised approach named Alleviating Semantic-level Shift (ASS), which can promote the distribution consistency from both global and local views. We apply our ASS to two domain adaptation tasks, from GTA5 to Cityscapes and from Synthia to Cityscapes. Extensive experiments demonstrate that: (1) ASS can significantly outperform the current unsupervised state-of-the-arts by employing a small number of annotated samples from the target domain; (2) ASS can beat the oracle model trained on the whole target dataset by over 3 points by augmenting the synthetic source data with annotated samples from the target domain without suffering from the prevalent problem of overfitting to the source domain.
AB - Utilizing synthetic data for semantic segmentation can significantly relieve human efforts in labelling pixel-level masks. A key challenge of this task is how to alleviate the data distribution discrepancy between the source and target domains, i.e. reducing domain shift. The common approach to this problem is to minimize the discrepancy between feature distributions from different domains through adversarial training. However, directly aligning the feature distribution globally cannot guarantee consistency from a local view (i.e. semantic-level). To tackle this issue, we propose a semi-supervised approach named Alleviating Semantic-level Shift (ASS), which can promote the distribution consistency from both global and local views. We apply our ASS to two domain adaptation tasks, from GTA5 to Cityscapes and from Synthia to Cityscapes. Extensive experiments demonstrate that: (1) ASS can significantly outperform the current unsupervised state-of-the-arts by employing a small number of annotated samples from the target domain; (2) ASS can beat the oracle model trained on the whole target dataset by over 3 points by augmenting the synthetic source data with annotated samples from the target domain without suffering from the prevalent problem of overfitting to the source domain.
UR - http://www.scopus.com/inward/record.url?scp=85090109245&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85090109245&partnerID=8YFLogxK
U2 - 10.1109/CVPRW50498.2020.00476
DO - 10.1109/CVPRW50498.2020.00476
M3 - Conference contribution
AN - SCOPUS:85090109245
T3 - IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops
SP - 4043
EP - 4047
BT - Proceedings - 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, CVPRW 2020
PB - IEEE Computer Society
Y2 - 14 June 2020 through 19 June 2020
ER -