OOD-CV: A Benchmark for Robustness to Out-of-Distribution Shifts of Individual Nuisances in Natural Images

TL;DR

This paper introduces OOD-CV, a comprehensive benchmark dataset for evaluating the robustness of vision models against various out-of-distribution nuisances like pose, shape, texture, and weather, across multiple tasks.

Contribution

The paper presents a new dataset, OOD-CV, for benchmarking robustness to individual nuisance factors in natural images, along with extensive baseline experiments highlighting current limitations.

Findings

Certain nuisance factors significantly degrade performance.

Existing robustness methods offer limited improvements.

No clear advantage of transformers over CNNs in robustness.

Abstract

Enhancing the robustness of vision algorithms in real-world scenarios is challenging. One reason is that existing robustness benchmarks are limited, as they either rely on synthetic data or ignore the effects of individual nuisance factors. We introduce OOD-CV, a benchmark dataset that includes out-of-distribution examples of 10 object categories in terms of pose, shape, texture, context and the weather conditions, and enables benchmarking models for image classification, object detection, and 3D pose estimation. In addition to this novel dataset, we contribute extensive experiments using popular baseline methods, which reveal that: 1. Some nuisance factors have a much stronger negative effect on the performance compared to others, also depending on the vision task. 2. Current approaches to enhance robustness have only marginal effects, and can even reduce robustness. 3. We do not observe significant differences between convolutional and transformer architectures. We believe our dataset provides a rich testbed to study robustness and will help push forward research in this area.