COREVQA: A Crowd Observation and Reasoning Entailment Visual Question Answering Benchmark
COREVQA: A Crowd Observation and Reasoning Entailment Visual Question Answering Benchmark
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.
Ishant Chintapatla、Kazuma Choji、Naaisha Agarwal、Andrew Lin、Hannah You、Charles Duong、Kevin Zhu、Sean O'Brien、Vasu Sharma
计算技术、计算机技术
Ishant Chintapatla,Kazuma Choji,Naaisha Agarwal,Andrew Lin,Hannah You,Charles Duong,Kevin Zhu,Sean O'Brien,Vasu Sharma.COREVQA: A Crowd Observation and Reasoning Entailment Visual Question Answering Benchmark[EB/OL].(2025-07-17)[2025-08-10].https://arxiv.org/abs/2507.13405.点此复制
评论