The paper reframes OOD detection as semantic anomaly detection, proposes hold-out-class benchmarks including fine-grained ImageNet subsets, and reports that rotation-prediction and CPC auxiliary tasks improve both anomaly detection and classification accuracy.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Detecting semantic anomalies
The paper reframes OOD detection as semantic anomaly detection, proposes hold-out-class benchmarks including fine-grained ImageNet subsets, and reports that rotation-prediction and CPC auxiliary tasks improve both anomaly detection and classification accuracy.