Abstract
Semi-supervised semantic segmentation (SS-SS) aims to mitigate the heavy annotation burden of dense pixel labeling by leveraging abundant unlabeled images alongside a small labeled set. While current teacher-student consistency regularization methods achieve strong results, they often overlook a critical challenge: the precise delineation of object boundaries. In this paper, we propose BoundMatch, a novel multi-task SS-SS framework that explicitly integrates semantic boundary detection into the consistency regularization pipeline. Our core mechanism, Boundary Consistency Regularized Multi-Task Learning (BCRM), enforces prediction agreement between teacher and student models on both segmentation masks and detailed semantic boundaries. To further enhance performance and sharpen contours, BoundMatch incorporates two lightweight fusion modules: Boundary-Semantic Fusion (BSF) injects learned boundary cues into the segmentation decoder, while Spatial Gradient Fusion (SGF) refines boundary predictions using mask gradients, leading to higher-quality boundary pseudo-labels. This framework is built upon SAMTH, a strong teacher-student baseline featuring a Harmonious Batch Normalization (HBN) update strategy for improved stability. Extensive experiments on diverse datasets including Cityscapes, BDD100K, SYNTHIA, ADE20K, and Pascal VOC show that BoundMatch achieves competitive performance against state-of-the-art methods while significantly improving boundary-specific evaluation metrics. We also demonstrate its effectiveness in realistic large-scale unlabeled data scenarios and on lightweight architectures designed for mobile deployment.
Abstract (translated)
半监督语义分割（SS-SS）旨在通过利用大量未标记的图像和少量已标注的数据集来减轻密集像素标签注释的工作负担。尽管当前的教师-学生一致性正则化方法取得了强大的成果，但它们往往忽视了一个关键挑战：精确地划分物体边界。在本文中，我们提出了一种名为BoundMatch的新颖多任务半监督语义分割框架，该框架明确将语义边界的检测集成到了一致性正则化的管道之中。我们的核心机制是边界一致性正则化多任务学习（BCRM），它要求教师和学生模型在分割掩码及详细的语义边界上达成预测的一致性。为了进一步提升性能并锐化轮廓，BoundMatch结合了两个轻量级融合模块：边界-语义融合（BSF）将学习到的边界线索注入到解码器中，而空间梯度融合（SGF）则利用掩码梯度细化边界预测，从而生成更高质量的边界伪标签。该框架基于SAMTH构建，这是一个强大的教师-学生基线模型，采用了和谐批量归一化（HBN）更新策略以提高稳定性。 在包括Cityscapes、BDD100K、SYNTHIA、ADE20K和Pascal VOC在内的多样化数据集上进行的广泛实验表明，BoundMatch取得了与最先进方法相当的表现，并显著提升了特定边界评价指标。此外，我们还展示了该框架在现实大规模未标注数据场景以及面向移动部署的轻量级架构上的有效性。
URL
https://arxiv.org/abs/2503.23519
