BenchReAD supplies a systematic retinal anomaly detection benchmark that shows DRA performs best overall yet drops on unseen cases, while NFM-DRA with added normal feature memory reaches new state-of-the-art results.
arXiv preprint arXiv:2404.04518 (2024)
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.CV 2verdicts
UNVERDICTED 2representative citing papers
AUCp selects inference models for unsupervised abnormality detection by computing AUC after labeling all test samples as positive, shown to outperform conventional metrics when normal training data is representative.
citing papers explorer
-
BenchReAD: A systematic benchmark for retinal anomaly detection
BenchReAD supplies a systematic retinal anomaly detection benchmark that shows DRA performs best overall yet drops on unseen cases, while NFM-DRA with added normal feature memory reaches new state-of-the-art results.
-
AUCp: Pseudo-AUC for Inference Model Selection with Unlabeled Validation Data in Abnormality Detection
AUCp selects inference models for unsupervised abnormality detection by computing AUC after labeling all test samples as positive, shown to outperform conventional metrics when normal training data is representative.