SOOD-ImageNet: a Large-Scale Dataset for Semantic Out-Of-Distribution Image Classification and Semantic Segmentation

TL;DR

SOOD-ImageNet is a large-scale dataset with 1.6 million images designed to improve out-of-distribution detection in computer vision, especially addressing semantic shift challenges in classification and segmentation tasks.

Contribution

The paper introduces SOOD-ImageNet, a novel large-scale dataset specifically targeting semantic shift in OOD detection, created with a new data engine and human verification.

Findings

Models trained on SOOD-ImageNet show improved OOD detection performance.

The dataset highlights the importance of semantic shift in OOD tasks.

Extensive evaluations demonstrate SOOD-ImageNet's utility for advancing OOD research.

Abstract

Out-of-Distribution (OOD) detection in computer vision is a crucial research area, with related benchmarks playing a vital role in assessing the generalizability of models and their applicability in real-world scenarios. However, existing OOD benchmarks in the literature suffer from two main limitations: (1) they often overlook semantic shift as a potential challenge, and (2) their scale is limited compared to the large datasets used to train modern models. To address these gaps, we introduce SOOD-ImageNet, a novel dataset comprising around 1.6M images across 56 classes, designed for common computer vision tasks such as image classification and semantic segmentation under OOD conditions, with a particular focus on the issue of semantic shift. We ensured the necessary scalability and quality by developing an innovative data engine that leverages the capabilities of modern vision-language models, complemented by accurate human checks. Through extensive training and evaluation of various models on SOOD-ImageNet, we showcase its potential to significantly advance OOD research in computer vision. The project page is available at https://github.com/bach05/SOODImageNet.git.