Domain Adaptation through Photorealistic Enhanced Images for Semantic Segmentation

Figure 2

Overview of the proposed framework for domain adaptation. The photorealistic enhanced images are generated by (a) image-level domain adaptation. In (b) interdomain adaptation, the adversarial learning represents that and are optimized by minimizing the segmentation loss and the adversarial loss . In (c) intradomain adaptation, the adversarial learning represents that and are optimized by using the intradomain segmentation loss and the adversarial loss .