PixOOD: Pixel-Level Out-of-Distribution Detection

TL;DR

PixOOD is a novel pixel-level out-of-distribution detection method that does not require anomalous training data and employs an online data condensation algorithm for robust modeling of in-distribution variability.

Contribution

The paper introduces PixOOD, a new dense prediction OOD detection algorithm that uses an online data condensation technique, avoiding traditional training biases and requiring no anomalous data.

Findings

Achieved state-of-the-art results on four datasets

Competitive performance on remaining datasets

Effective modeling of intra-class variability

Abstract

We propose a dense image prediction out-of-distribution detection algorithm, called PixOOD, which does not require training on samples of anomalous data and is not designed for a specific application which avoids traditional training biases. In order to model the complex intra-class variability of the in-distribution data at the pixel level, we propose an online data condensation algorithm which is more robust than standard K-means and is easily trainable through SGD. We evaluate PixOOD on a wide range of problems. It achieved state-of-the-art results on four out of seven datasets, while being competitive on the rest. The source code is available at https://github.com/vojirt/PixOOD.