PulseAugur
中
实时 19:53:03
English(EN) GFSR: Geometric Fidelity and Spatial Refinement for Reliable Lane Detection

SAM通过像素级标注增强自动驾驶数据集

作者 PulseAugur 编辑部 · [5 个来源] ·

研究人员开发了一种使用Segment Anything Model (SAM) 的新流程，为以前只有边界框的自动驾驶数据集生成密集、像素级的标注。这种基于SAM的方法已应用于Zenseact Open Dataset (ZOD)，生成了超过10万帧的标注，其中精选的2300帧子集显示了36%的接受率。在基于Transformer的CLFT和基于CNN的DeepLabV3+架构上的评估达到了48.1%的mIoU，而专门的模型对稀有类别显示出潜力。该流程在Iseauto平台上得到了进一步验证，达到了77.5%的mIoU，并展示了有效的跨传感器迁移学习。 AI

影响 提高了自动驾驶研究数据集的质量和效用，有可能加速更强大的感知系统的开发。

排序理由 该集群包含多篇arXiv论文，详细介绍了改进自动驾驶数据集和模型的研究。

在 arXiv cs.CV 阅读 →

AI 生成摘要 · Google Gemini · 来自 5 个来源。 我们如何撰写摘要 →

SAM通过像素级标注增强自动驾驶数据集

报道来源 [5]

  1. arXiv cs.CV TIER_1 English(EN) · Toomas Tahves, Mauro Bellone, Junyi Gu, Raivo Sell ·

    SAM-Enhanced Segmentation on Road Datasets: Balancing Critical Classes in Autonomous Driving

    arXiv:2605.28136v1 Announce Type: new Abstract: Dense semantic segmentation is essential for autonomous driving, yet many multi-modal datasets lack pixel-level annotations. The Zenseact Open Dataset (ZOD) provides rich multi-sensor data but only bounding-box labels, limiting its …

  2. arXiv cs.CV TIER_1 English(EN) · Raivo Sell ·

    SAM-Enhanced Segmentation on Road Datasets: Balancing Critical Classes in Autonomous Driving

    Dense semantic segmentation is essential for autonomous driving, yet many multi-modal datasets lack pixel-level annotations. The Zenseact Open Dataset (ZOD) provides rich multi-sensor data but only bounding-box labels, limiting its use for segmentation research. Our primary contr…

  3. arXiv cs.CV TIER_1 English(EN) · J\"org Gamerdinger, Sven Teufel, Oliver Bringmann ·

    Datasets for Lane Detection in Autonomous Driving: A Comprehensive Review

    arXiv:2504.08540v2 Announce Type: replace Abstract: Accurate lane detection is essential for automated driving, enabling safe and reliable vehicle navigation across a variety of road scenarios. Numerous datasets have been introduced to support the development and evaluation of la…

  4. arXiv cs.CV TIER_1 English(EN) · Tiancheng Wang, Zhaolu Ding, Richeng Xu, Tianhui Zheng, Hui Liu, Hanyu Xuan, Zhiliang Wu, Guanghui Yue ·

    GFSR: Geometric Fidelity and Spatial Refinement for Reliable Lane Detection

    arXiv:2605.23327v1 Announce Type: new Abstract: Lane detection stands as a crucial perception task in autonomous driving and advanced driver assistance systems. However, existing methods still degrade in complex real scenarios due to two major limitations. First, classification c…

  5. arXiv cs.CV TIER_1 English(EN) · Guanghui Yue ·

    GFSR: Geometric Fidelity and Spatial Refinement for Reliable Lane Detection

    Lane detection stands as a crucial perception task in autonomous driving and advanced driver assistance systems. However, existing methods still degrade in complex real scenarios due to two major limitations. First, classification confidence only characterizes the categorical exi…

相关实体

相关话题