Go to ICLR 2025 Conference homepageBalanced Learning for Domain Adaptive Semantic Segmentation
Unsupervised domain adaptation (UDA) for semantic segmentation aims to transfer knowledge from a labeled source domain to an unlabeled target domain, improving model performance on the target dataset without additional annotations. Despite the effectiveness of self-training techniques in UDA, they struggle to learn each class in a balanced manner due to inherent class imbalance and distribution shift in both data and label space between domains. To address this issue, we propose Balanced Learning for Domain Adaptation (BLDA), a novel approach to directly assess and alleviate class bias without requiring prior knowledge about the distribution shift between domains. First, we identify over-predicted and under-predicted classes by analyzing the distribution of predicted logits. Subsequently, we introduce a post-hoc approach to align the positive and negative logits distributions across different classes using anchor distributions and cumulative density functions. To further consider the network's need to generate unbiased pseudo-labels during self-training, we couple Gaussian mixture models to estimate logits distributions online and incorporate logits correction terms into the loss function. Moreover, we leverage the resulting cumulative density as domain-shared structural knowledge to connect the source and target domains. Extensive experiments on two standard UDA semantic segmentation benchmarks demonstrate that BLDA consistently improves performance, especially for under-predicted classes, when integrated into existing methods. Our work highlights the importance of balanced learning in UDA and effectively mitigates class bias in domain adaptive semantic segmentation.