Calibrated Out-of-Distribution Detection with a Generic Representation

TL;DR

This paper introduces GROOD, a simple and efficient out-of-distribution detection method that leverages pre-trained representations and formulates detection as a Neyman-Pearson task, achieving state-of-the-art results.

Contribution

Proposes GROOD, a novel OOD detection approach using generic pre-trained features and Neyman-Pearson formulation, requiring minimal training and offering calibrated scores.

Findings

Achieves near-perfect performance on several benchmarks.

Outperforms existing OOD detection methods.

Requires only trivial adaptation for specific tasks.

Abstract

Out-of-distribution detection is a common issue in deploying vision models in practice and solving it is an essential building block in safety critical applications. Most of the existing OOD detection solutions focus on improving the OOD robustness of a classification model trained exclusively on in-distribution (ID) data. In this work, we take a different approach and propose to leverage generic pre-trained representation. We propose a novel OOD method, called GROOD, that formulates the OOD detection as a Neyman-Pearson task with well calibrated scores and which achieves excellent performance, predicated by the use of a good generic representation. Only a trivial training process is required for adapting GROOD to a particular problem. The method is simple, general, efficient, calibrated and with only a few hyper-parameters. The method achieves state-of-the-art performance on a number of OOD benchmarks, reaching near perfect performance on several of them. The source code is available at https://github.com/vojirt/GROOD.