Multi-Source Expert Scoring for zero-shot visual anomaly classification and segmentation

Hui Gao

doi:10.1016/j.cviu.2026.104794

Multi-Source Expert Scoring for zero-shot visual anomaly classification and segmentation

2026-04-30

0

PRE

AI

DOI：10.1016/j.cviu.2026.104794

Original

Original request for help

Save

Abstract

En 中文

• MS-ExSc is a zero-shot anomaly detection method that outperforms SOTA techniques. • Multi-Source Feature Extraction offers a more comprehensive representation of data. • The Expert Scoring Mechanism mitigates the influence of anomalous samples. • The Cascade Fusion Strategy improves the performance of anomaly evaluation.

Keywords:

zero-shot anomaly detection

multi-source feature extraction

expert scoring mechanism

cascade fusion strategy

Multi-Source Expert Scoring for zero-shot visual anomaly classification and segmentation

Abstract

Journal

Organization