TY - GEN
T1 - SEMI-DERAINGAN
T2 - 2021 IEEE International Conference on Multimedia and Expo, ICME 2021
AU - Wei, Yanyan
AU - Zhang, Zhao
AU - Wang, Yang
AU - Zhang, Haijun
AU - Zhao, Mingbo
AU - Xu, Mingliang
AU - Wang, Meng
PY - 2021/7
Y1 - 2021/7
N2 - Although supervised single image deraining (SID) have obtained impressive results, they still cannot obtain satisfactory results on real images for the weak generalization of rain removal capacity. In this paper, we mainly discuss the semi-supervised SID and propose a new GAN-based deraining network called Semi-DerainGAN, which can use both synthetic and real data in a uniform network based on two supervised and unsupervised processes. For this task, a semi-supervised rain streak learner termed SSRML sharing the same parameters of both processes is derived, which makes the real images contribute more rain streak information, so that the resulted model has a strong generalization power to the real SID task. We also contribute a new real-world rain image dataset called Real200 to alleviate the difference between both synthetic and real image domains. Extensive results on public datasets show that our model can obtain competitive results, especially on the real rain images.
AB - Although supervised single image deraining (SID) have obtained impressive results, they still cannot obtain satisfactory results on real images for the weak generalization of rain removal capacity. In this paper, we mainly discuss the semi-supervised SID and propose a new GAN-based deraining network called Semi-DerainGAN, which can use both synthetic and real data in a uniform network based on two supervised and unsupervised processes. For this task, a semi-supervised rain streak learner termed SSRML sharing the same parameters of both processes is derived, which makes the real images contribute more rain streak information, so that the resulted model has a strong generalization power to the real SID task. We also contribute a new real-world rain image dataset called Real200 to alleviate the difference between both synthetic and real image domains. Extensive results on public datasets show that our model can obtain competitive results, especially on the real rain images.
KW - dataset
KW - rain removal
KW - Semi-supervised learning
KW - Single image deraining
UR - http://www.scopus.com/inward/record.url?scp=85126434582&partnerID=8YFLogxK
UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-85126434582&origin=recordpage
U2 - 10.1109/ICME51207.2021.9428285
DO - 10.1109/ICME51207.2021.9428285
M3 - RGC 32 - Refereed conference paper (with host publication)
T3 - Proceedings - IEEE International Conference on Multimedia and Expo
BT - 2021 IEEE International Conference on Multimedia and Expo (ICME)
PB - IEEE
Y2 - 5 July 2021 through 9 July 2021
ER -