【深度学习】【CVPR2021】ProDA
Prototypical Pseudo Label Denoising and Target Structure Learning for Domain Adaptive Semantic Segmentation摘要intro相关工作Unsupervised domain adaptationUnsupervised representation learningLearning from noisy labels方法摘要self-training: 伪标签训练目标域伪标签的缺点: 由于源域和目标











