All about Structure: Adapting Structural Information across Domains for Boosting Semantic Segmentation
arXiv:1903.12212
Abstract
In this paper we tackle the problem of unsupervised domain adaptation for the task of semantic segmentation, where we attempt to transfer the knowledge learned upon synthetic datasets with ground-truth labels to real-world images without any annotation. With the hypothesis that the structural content of images is the most informative and decisive factor to semantic segmentation and can be readily shared across domains, we propose a Domain Invariant Structure Extraction (DISE) framework to disentangle images into domain-invariant structure and domain-specific texture representations, which can further realize image-translation across domains and enable label transfer to improve segmentation performance. Extensive experiments verify the effectiveness of our proposed DISE model and demonstrate its superiority over several state-of-the-art approaches.
References in corpus (16)
- Very Deep Convolutional Networks for Large-Scale Image Recognition
- Deep Residual Learning for Image Recognition
- Unsupervised Domain Adaptation by Backpropagation
- Multi-Scale Context Aggregation by Dilated Convolutions
- The Cityscapes Dataset for Semantic Urban Scene Understanding
- Fully Convolutional Networks for Semantic Segmentation
- DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs
- CyCADA: Cycle-Consistent Adversarial Domain Adaptation
- Adversarial Discriminative Domain Adaptation
- Deep CORAL: Correlation Alignment for Deep Domain Adaptation
- Pyramid Scene Parsing Network
- DRIT++: Diverse Image-to-Image Translation via Disentangled Representations
- Least Squares Generative Adversarial Networks
- Playing for Data: Ground Truth from Computer Games
- Effective Use of Synthetic Data for Urban Scene Semantic Segmentation
- Penalizing Top Performers: Conservative Loss for Semantic Segmentation Adaptation