[论文] Modulate-and-Map: Crossmodal Feature Mapping with Cross-View Modulatio...

## 论文概要 **研究领域**: CV **作者**: Alex Costanzino, Pierluigi Zama Ramirez, Giuseppe Lisanti **发布时间**: 2025-04-01 **arXiv**: [2504.01262](https://arxiv.org/abs/2504.01262) ## 中文摘要 我们提出了 ModMap，一个原生的多视角和多模态3D异常检测与分割框架。与独立处理视角的现有方法不同，我们的方法从跨模态特征映射范式中汲取灵感，学习跨模态和视角映射特征，同时通过特征级调制显式建模视角相关关系。我们引入了一种利用所有可能视角组合的跨视角训练策略，通过多视角集成和聚合实现有效的异常评分。为了处理高分辨率3D数据，我们训练并公开发布了一个针对工业数据集定制的基础深度编码器。在 SiM3D（一个引入了首个多视角多模态3D异常检测和分割设置的最新基准）上的实验表明，ModMap 以大幅优势超越先前方法，达到了最先进的性能。 ## 原文摘要 We present ModMap, a natively multiview and multimodal framework for 3D anomaly detection and segmentation. Unlike existing methods that process views independently, our method draws inspiration from the crossmodal feature mapping paradigm to learn to map features across both modalities and views, while explicitly modelling view-dependent relationships through feature-wise modulation. We introduce a cross-view training strategy that leverages all possible view combinations, enabling effective anomaly scoring through multiview ensembling and aggregation. To process high-resolution 3D data, we train and publicly release a foundational depth encoder tailored to industrial datasets. Experiments on SiM3D, a recent benchmark that introduces the first multiview and multimodal setup for 3D anomaly... --- *自动采集于 2026-04-04* #论文 #arXiv #CV #小凯