arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2602.10052cs.CV
收藏

时空注意力用于自动驾驶中的视频语义分割一致性

Spatio-Temporal Attention for Consistent Video Semantic Segmentation in Automated Driving

1 Heinrich-Heine-University D \"u sseldorf, Department of Computer Science, D \"u sseldorf, Germany

Serin Varghese, Kevin Ross, Fabian Hueger, Kira Maag

AI总结:

本文提出时空注意力机制,通过多帧上下文增强视频语义分割的时序特征表示,提升动态场景的准确性和稳定性,实验显示在Cityscapes和BDD100k数据集上显著提升时序一致性指标和mIoU。

摘要:

深度神经网络,特别是基于变换器的架构,在环境感知的语义分割中取得了显著成功。然而,现有模型独立处理视频帧,未能利用时间一致性,这可能显著提高动态场景的准确性和稳定性。本文提出了一种时空注意力(STA)机制,扩展变换器注意力块以纳入多帧上下文,从而为视频语义分割提供稳健的时间特征表示。我们的方法修改标准自注意力以处理时空特征序列,同时保持计算效率,并对现有架构进行最小修改。STA在各种变换器架构中具有广泛的应用性,并在轻量级和大尺度模型中均保持有效性。在Cityscapes和BDD100k数据集上的全面评估显示,STA在时序一致性指标上提高了9.20个百分点,在mIoU上比单帧基线提高了1.76个百分点。这些结果表明STA是视频语义分割应用的有效架构增强。

英文摘要:

Deep neural networks, especially transformer-based architectures, have achieved remarkable success in semantic segmentation for environmental perception. However, existing models process video frames independently, thus failing to leverage temporal consistency, which could significantly improve both accuracy and stability in dynamic scenes. In this work, we propose a Spatio-Temporal Attention (STA) mechanism that extends transformer attention blocks to incorporate multi-frame context, enabling robust temporal feature representations for video semantic segmentation. Our approach modifies standard self-attention to process spatio-temporal feature sequences while maintaining computational efficiency and requiring minimal changes to existing architectures. STA demonstrates broad applicability across diverse transformer architectures and remains effective across both lightweight and larger-scale models. A comprehensive evaluation on the Cityscapes and BDD100k datasets shows substantial improvements of 9.20 percentage points in temporal consistency metrics and up to 1.76 percentage points in mean intersection over union compared to single-frame baselines. These results demonstrate STA as an effective architectural enhancement for video-based semantic segmentation applications.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑