Learning from synthetic data: Addressing domain shift for semantic segmentation.
- 1 minLearning from synthetic data: Addressing domain shift for semantic segmentation.
[CVPR2018] Sankaranarayanan, S., Balaji, Y., Jain, A., Nam Lim, S., & Chellappa, R. (UMIACS, University of Maryland, College Park, MD, GE Global Research, Niskayuna, NY and Avitas Systems, GE Venture, Boston MA). Learning from synthetic data: Addressing domain shift for semantic segmentation. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 3752-3761).
简介
这篇文章介绍的通过重建合成真实图像去解决域适应问题。
其应用示意图如下：
图1 思路示意图
相关工作
对抗方法。（7,2,28,19,14,32）
Ganin, Y., & Lempitsky, V. (2015, June). Unsupervised domain adaptation by backpropagation. In International conference on machine learning (pp. 1180-1189). PMLR.
Bousmalis, K., Silberman, N., Dohan, D., Erhan, D., & Krishnan, D. (2017). Unsupervised pixel-level domain adaptation with generative adversarial networks. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 3722-3731).
Sankaranarayanan, S., Balaji, Y., Castillo, C. D., & Chellappa, R. (2018). Generate to adapt: Aligning domains using generative adversarial networks. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 8503-8512).
Liu, M. Y., & Tuzel, O. (2016). Coupled generative adversarial networks. In Advances in neural information processing systems (pp. 469-477).
Hoffman, J., Wang, D., Yu, F., & Darrell, T. (2016). Fcns in the wild: Pixel-level adversarial and constraint-based adaptation. arXiv preprint arXiv:1612.02649.
Zhang, Y., David, P., & Gong, B. (2017). Curriculum domain adaptation for semantic segmentation of urban scenes. In Proceedings of the IEEE International Conference on Computer Vision (pp. 2020-2030).
方法
本文介绍的网络结构示意图如下：
图2 网络结构示意图
网络包括四个部分，特征提取（F网络），生成网络（G网络），分类网络（C网络）以及判别网络（D网络）。控制的Loss如下表所示：
实验和结果
实验数据集包括SYNTHIA->Cityscapes,以及GTA5->Cityscapes。 实验结果如下： 