Search papers, labs, and topics across Lattice.
The paper introduces ST-Seg, a novel framework to address distribution shifts in semantic segmentation for off-road autonomous navigation by expanding the source distribution. ST-Seg uses style expansion (SE) to generate diverse realistic styles, augmenting the limited style information of the source domain, and texture regularization (TR) to stabilize local texture representation. Experiments across distribution-shifted target domains demonstrate ST-Seg's effectiveness, showing substantial improvements over existing methods in semantic label prediction accuracy.
Style augmentation and texture regularization can significantly improve semantic segmentation robustness in off-road environments facing distribution shifts.
Semantic segmentation is crucial for autonomous navigation in off-road environments, enabling precise classification of surroundings to identify traversable regions. However, distinctive factors inherent to off-road conditions, such as source-target domain discrepancies and sensor corruption from rough terrain, can result in distribution shifts that alter the data differently from the trained conditions. This often leads to inaccurate semantic label predictions and subsequent failures in navigation tasks. To address this, we propose ST-Seg, a novel framework that expands the source distribution through style expansion (SE) and texture regularization (TR). Unlike prior methods that implicitly apply generalization within a fixed source distribution, ST-Seg offers an intuitive approach for distribution shift. Specifically, SE broadens domain coverage by generating diverse realistic styles, augmenting the limited style information of the source domain. TR stabilizes local texture representation affected by style-augmented learning through a deep texture manifold. Experiments across various distribution-shifted target domains demonstrate the effectiveness of ST-Seg, with substantial improvements over existing methods. These results highlight the robustness of ST-Seg, enhancing the real-world applicability of semantic segmentation for off-road navigation.
