Multimodal Industrial Anomaly Detection by Crossmodal Feature Mapping

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Costanzino, Alex, Ramirez, Pierluigi Zama, Lisanti, Giuseppe, Di Stefano, Luigi
Formato: Preprint
Publicado: 2023
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866910516930150400
author Costanzino, Alex
Ramirez, Pierluigi Zama
Lisanti, Giuseppe
Di Stefano, Luigi
author_facet Costanzino, Alex
Ramirez, Pierluigi Zama
Lisanti, Giuseppe
Di Stefano, Luigi
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2312_04521
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Multimodal Industrial Anomaly Detection by Crossmodal Feature Mapping
Costanzino, Alex
Ramirez, Pierluigi Zama
Lisanti, Giuseppe
Di Stefano, Luigi
Computer Vision and Pattern Recognition
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.
title Multimodal Industrial Anomaly Detection by Crossmodal Feature Mapping
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2312.04521