Diverse, Difficult, and Odd Instances (D2O): A New Test Set for Object Classification

TL;DR

The paper introduces D2O, a new diverse and challenging test set for object classification, designed to better evaluate model generalization and reveal weaknesses in current AI systems.

Contribution

D2O is a novel test set with diverse, real-world images that differ from existing datasets, highlighting limitations of current models and APIs in object recognition.

Findings

Models achieve around 60% accuracy on D2O, much lower than on ImageNet.

Popular vision APIs perform poorly on D2O categories like faces, cars, and cats.

D2O's varied difficulty levels make it a strong predictor of model performance.

Abstract

Test sets are an integral part of evaluating models and gauging progress in object recognition, and more broadly in computer vision and AI. Existing test sets for object recognition, however, suffer from shortcomings such as bias towards the ImageNet characteristics and idiosyncrasies (e.g., ImageNet-V2), being limited to certain types of stimuli (e.g., indoor scenes in ObjectNet), and underestimating the model performance (e.g., ImageNet-A). To mitigate these problems, we introduce a new test set, called D2O, which is sufficiently different from existing test sets. Images are a mix of generated images as well as images crawled from the web. They are diverse, unmodified, and representative of real-world scenarios and cause state-of-the-art models to misclassify them with high confidence. To emphasize generalization, our dataset by design does not come paired with a training set. It contains 8,060 images spread across 36 categories, out of which 29 appear in ImageNet. The best Top-1 accuracy on our dataset is around 60% which is much lower than 91% best Top-1 accuracy on ImageNet. We find that popular vision APIs perform very poorly in detecting objects over D2O categories such as ``faces'', ``cars'', and ``cats''. Our dataset also comes with a ``miscellaneous'' category, over which we test the image tagging models. Overall, our investigations demonstrate that the D2O test set contain a mix of images with varied levels of difficulty and is predictive of the average-case performance of models. It can challenge object recognition models for years to come and can spur more research in this fundamental area.