# COREVQA: A Crowd Observation and Reasoning Entailment Visual Question Answering Benchmark

Source: [COREVQA: A Crowd Observation and Reasoning Entailment Visual Question Answering Benchmark](<https://www.sota2.com/research/paper/68fa7a181e209131cea811ee>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-10-26T05:12:23.158Z

Authors: Ishant Chintapatla, Kazuma Choji, Naaisha Agarwal, Andrew Lin, Hannah You, Charles Duong, Kevin Zhu, Sean O'Brien, Vasu Sharma

Published: 2025-07-17T04:47:47.000Z

## Abstract

Recently, many benchmarks and datasets have been developed to evaluate Vision-Language Models (VLMs) using visual question answering (VQA) pairs, and models have shown significant accuracy improvements. However, these benchmarks rarely test the model's ability to accurately complete visual entailment, for instance, accepting or refuting a hypothesis based on the image. To address this, we propose COREVQA (Crowd Observations and Reasoning Entailment), a benchmark of 5608 image and synthetically generated true/false statement pairs, with images derived from the CrowdHuman dataset, to provoke visual entailment reasoning on challenging crowded images. Our results show that even the top-performing VLMs achieve accuracy below 80%, with other models performing substantially worse (39.98%-69.95%). This significant performance gap reveals key limitations in VLMs' ability to reason over certain types of image-question pairs in crowded scenes.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2507.13405>)

[PDF](<https://arxiv.org/pdf/2507.13405.pdf>)