Skip to content
Hot eventLive

TopoEnhance:自动驾驶场景拓扑增强框架

1 reports1 sources3 hr ago updated

Get the story

AI overview

2026年10月9日,研究者在 arXiv(Computer Vision)发表论文,提出面向驾驶场景拓扑的可靠增强框架 TopoEnhance。该框架将拓扑增强建模为去噪重建过程,从随机损坏的真值图中恢复结构一致性,以修正阈值化连续拓扑分数带来的误连与漏连问题。实验显示,TopoEnhance 在不同基线上同时提升连续指标 TOP score 与离散连通性指标 Topology Jaccard Similarity(TJS),且无需重新训练即可适配多种现有方法。目前报道仅涉及该框架的方法与实验结果,未见后续验证或应用进展。

Generated from reports · updated 3 hr ago

Timeline

Follow the coverage from different angles.

Oct 9, 2026
  1. arXiv · Computer Vision
    TopoEnhance:面向驾驶场景拓扑的可靠增强框架

    研究者提出驾驶场景拓扑增强框架 TopoEnhance,把拓扑增强建模为去噪重建过程,从随机损坏的真值图中恢复结构一致性,以修正阈值化连续拓扑分数带来的误连与漏连。实验显示,TopoEnhance 在不同基线上同时提升连续指标 TOP score 与离散连通性指标 Topology Jaccard Similarity(TJS),无需重新训练即可适配多种现有方法。

Heat trend

Current heat 9·Comparable peak 10(Oct 9)·Comparable change over 24 hours –

02.557.510Oct9Oct9Oct9Oct9

The trend compares only the same participants observed continuously; its range may be smaller than the current heat count. Move or click on the chart to inspect hourly heat; use the left and right arrow keys to switch.