UniSLAD unifies structural and logical anomaly detection via dual CNN-Transformer features, Mahalanobis memory banks, and LUM/PMP pooling, reporting 99.4% and 93.1% on two industrial benchmarks.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
UniSLAD: A Unified Framework for Structural and Logical Industrial Visual Anomaly Detection
UniSLAD unifies structural and logical anomaly detection via dual CNN-Transformer features, Mahalanobis memory banks, and LUM/PMP pooling, reporting 99.4% and 93.1% on two industrial benchmarks.