标题： 精细的空间-时间感知用于气体泄漏分割 显示英文标题

标题： Fine-grained spatial-temporal perception for gas leak segmentation

摘要： 气体泄漏对人类健康和环境构成重大风险。尽管长期以来一直存在担忧，但由于泄漏的隐蔽性和随机形状，现有的方法有限，无法高效且准确地检测和分割泄漏。在本文中，我们提出了一种细粒度时空感知（FGSTP）算法用于气体泄漏分割。FGSTP捕获帧间的关键运动线索，并将其与精炼的目标特征集成在一个端到端网络中。具体来说，我们首先构建一个相关体积以捕捉连续帧之间的运动信息。然后，细粒度感知使用先前的输出逐步细化目标级特征。最后，采用解码器优化边界分割。由于没有高度精确标记的数据集用于气体泄漏分割，我们手动标注了一个气体泄漏视频数据集GasVid。实验结果显示，在GasVid上的结果表明我们的模型在分割非刚性物体（如气体泄漏）方面表现出色，与其他最先进的（SOTA）模型相比，生成了最准确的掩膜。 显示英文摘要

摘要： Gas leaks pose significant risks to human health and the environment. Despite long-standing concerns, there are limited methods that can efficiently and accurately detect and segment leaks due to their concealed appearance and random shapes. In this paper, we propose a Fine-grained Spatial-Temporal Perception (FGSTP) algorithm for gas leak segmentation. FGSTP captures critical motion clues across frames and integrates them with refined object features in an end-to-end network. Specifically, we first construct a correlation volume to capture motion information between consecutive frames. Then, the fine-grained perception progressively refines the object-level features using previous outputs. Finally, a decoder is employed to optimize boundary segmentation. Because there is no highly precise labeled dataset for gas leak segmentation, we manually label a gas leak video dataset, GasVid. Experimental results on GasVid demonstrate that our model excels in segmenting non-rigid objects such as gas leaks, generating the most accurate mask compared to other state-of-the-art (SOTA) models.