M2AD, a large-scale benchmark with 120 view-illumination configurations per object, shows that state-of-the-art visual anomaly detection methods drop markedly when viewpoint and lighting vary together.
Pad: A dataset and benchmark for pose-agnostic anomaly detection
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Visual Anomaly Detection under Complex View-Illumination Interplay: A Large-Scale Benchmark
M2AD, a large-scale benchmark with 120 view-illumination configurations per object, shows that state-of-the-art visual anomaly detection methods drop markedly when viewpoint and lighting vary together.