A Study of Commonsense Reasoning over Visual Object Properties

TL;DR

This paper introduces a systematic framework and benchmarks for evaluating vision-language models' reasoning over object properties, revealing significant gaps compared to human performance especially in complex and counterfactual reasoning tasks.

Contribution

It presents a new evaluation framework and benchmark datasets for assessing VLMs' reasoning abilities over object properties across multiple reasoning levels and image types.

Findings

VLMs perform below 40% in counting accuracy

VLMs achieve below 70% in comparison accuracy

Models struggle with photographic images and counterfactual reasoning

Abstract

Inspired by human categorization, object property reasoning involves identifying and recognizing low-level details and higher-level abstractions. While current visual question answering (VQA) studies consider multiple object properties, such as size, they typically blend perception and reasoning and lack representativeness in terms of reasoning and image categories, making it unclear whether and how vision-language models (VLMs) abstract and reason over depicted objects. To this end, we introduce a systematic evaluation framework comprising images of three representative types, three reasoning levels of increasing complexity, and four object property dimensions, informed by prior work on common sense. We develop a procedure to instantiate this framework in two VQA object reasoning benchmarks: OPTICS-CNT, comprising 360 images paired with 1,080 multi-level, count-based questions, and OPTICS-CMP, with 2.1k comparison questions. Experiments with 12 state-of-the-art VLMs in zero-shot settings reveal significant limitations relative to humans, with the best-performing model achieving below 40% counting and 70% comparison accuracy. VLMs struggle particularly with photographic images, counterfactual reasoning, physical and functional properties, and higher counts. We make the OPTICS benchmark data and code available to support future work on scalable benchmarking methods, generalized annotation guidelines, and advanced reasoning VLMs.