发表机构
Heinrich-Heine-University; CARIAD SE（海因里希·海涅大学; CARIAD SE）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
提出一种免训练的OOD分割方法，利用基础分割模型置信度并结合单目深度几何信息，在SegmentMeIfYouCan基准上表现强劲，无需微调或异常数据。
AI 中文摘要
在开放世界场景中运行的自动驾驶车辆不可避免地会遇到先前未知的物体，例如外来动物或散落货物。因此，可靠地检测和分割这些分布外（OOD）物体对于安全的环境理解和决策至关重要。现有的大多数方法需要访问OOD训练样本、重新训练分割骨干网络，或使用专门的辅助架构，这限制了它们的实际适用性。我们提出了一种免训练方法，该方法直接从基础分割模型的置信度预测中推导出密集的OOD分数，无需任何任务特定的微调或访问异常数据。为了提高OOD分割的鲁棒性，我们将单目深度估计的几何信息纳入决策过程，为基于不确定性的预测提供互补线索。我们在SegmentMeIfYouCan基准上评估了所提出的方法，并额外评估了其在视频序列中OOD跟踪的性能，这反映了真实世界感知系统的时间特性。该方法在道路中心基准上表现强劲。
英文摘要
Autonomous vehicles operating in open-world scenarios are inevitably confronted with previously unknown objects, such as exotic animals or loose cargo. The reliable detection and segmentation of these out-of-distribution (OOD) objects is therefore crucial for a safe understanding of the environment and decision-making. Most existing approaches require access to OOD training samples, retraining of the segmentation backbone, or dedicated auxiliary architectures, limiting their practical applicability. We propose a training-free method that derives dense OOD scores directly from the confidence predictions of a foundation segmentation model, without any task-specific fine-tuning or access to anomalous data. To improve the robustness of our OOD segmentation, geometric information from monocular depth estimation is incorporated into the decision process, providing complementary cues to uncertainty-based predictions. We evaluate the proposed method on the SegmentMeIfYouCan benchmark and additionally assess its performance on OOD tracking in video sequences, reflecting the temporal nature of real-world perception systems. The method performs strongly on road-centered benchmarks.
