Open-vocabulary Attribute Detection

TL;DR

This paper introduces the OVAD benchmark for open-vocabulary object attribute detection, providing a large, annotated dataset and baseline methods to evaluate vision-language models' ability to recognize object attributes in a zero-shot setting.

Contribution

The paper presents the first benchmark and dataset for open-vocabulary object attribute detection, enabling systematic evaluation of models' attribute recognition capabilities.

Findings

Baseline methods show limited attribute detection performance.

The benchmark reveals strengths and weaknesses of current vision-language models.

Open-vocabulary attribute detection remains a challenging task.

Abstract

Vision-language modeling has enabled open-vocabulary tasks where predictions can be queried using any text prompt in a zero-shot manner. Existing open-vocabulary tasks focus on object classes, whereas research on object attributes is limited due to the lack of a reliable attribute-focused evaluation benchmark. This paper introduces the Open-Vocabulary Attribute Detection (OVAD) task and the corresponding OVAD benchmark. The objective of the novel task and benchmark is to probe object-level attribute information learned by vision-language models. To this end, we created a clean and densely annotated test set covering 117 attribute classes on the 80 object classes of MS COCO. It includes positive and negative annotations, which enables open-vocabulary evaluation. Overall, the benchmark consists of 1.4 million annotations. For reference, we provide a first baseline method for open-vocabulary attribute detection. Moreover, we demonstrate the benchmark's value by studying the attribute detection performance of several foundation models. Project page https://ovad-benchmark.github.io