The paper explores the industrial multimodal Anomaly Detection (AD) task,
which exploits point clouds and RGB images to localize anomalies. We introduce
a novel light and fast framework that learns to map features from one modality
to the other on nominal samples. At test time, anomalies are detected by
pinpointing inconsistencies between observed and mapped features. Extensive
experiments show that our approach achieves state-of-the-art detection and
segmentation performance in both the standard and few-shot settings on the
MVTec 3D-AD dataset while achieving faster inference and occupying less memory
than previous multimodal AD methods. Moreover, we propose a layer-pruning
technique to improve memory and time efficiency with a marginal sacrifice in
performance.

这篇论文探讨了工业多模态异常检测任务，利用点云和 RGB 图像来定位异常。我们介绍了一种新颖且快速的框架，它学习将一个模态中的特征映射到另一个模态的正常样本上。在测试时，通过定位观察到的特征和映射特征之间的不一致性来检测异常。大量实验证明，我们的方法在 MVTec 3D-AD 数据集的标准和少样本设置中实现了最先进的检测和分割性能，同时实现了更快的推理速度和更低的内存占用，优于先前的多模态异常检测方法。此外，我们提出一种层裁剪技术，在性能上略有损失的同时提高了内存和时间效率。