HRScene: How Far Are VLMs from Effective High-Resolution Image Understanding?
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866913811981664256 |
|---|---|
| author | Zhang, Yusen Zheng, Wenliang Madasu, Aashrith Shi, Peng Kamoi, Ryo Zhou, Hao Zou, Zhuoyang Zhao, Shu Das, Sarkar Snigdha Sarathi Gupta, Vipul Lu, Xiaoxin Zhang, Nan Zhang, Ranran Haoran Iyer, Avitej Lou, Renze Yin, Wenpeng Zhang, Rui |
| author_facet | Zhang, Yusen Zheng, Wenliang Madasu, Aashrith Shi, Peng Kamoi, Ryo Zhou, Hao Zou, Zhuoyang Zhao, Shu Das, Sarkar Snigdha Sarathi Gupta, Vipul Lu, Xiaoxin Zhang, Nan Zhang, Ranran Haoran Iyer, Avitej Lou, Renze Yin, Wenpeng Zhang, Rui |
| contents | High-resolution image (HRI) understanding aims to process images with a large number of pixels, such as pathological images and agricultural aerial images, both of which can exceed 1 million pixels. Vision Large Language Models (VLMs) can allegedly handle HRIs, however, there is a lack of a comprehensive benchmark for VLMs to evaluate HRI understanding. To address this gap, we introduce HRScene, a novel unified benchmark for HRI understanding with rich scenes. HRScene incorporates 25 real-world datasets and 2 synthetic diagnostic datasets with resolutions ranging from 1,024 $\times$ 1,024 to 35,503 $\times$ 26,627. HRScene is collected and re-annotated by 10 graduate-level annotators, covering 25 scenarios, ranging from microscopic to radiology images, street views, long-range pictures, and telescope images. It includes HRIs of real-world objects, scanned documents, and composite multi-image. The two diagnostic evaluation datasets are synthesized by combining the target image with the gold answer and distracting images in different orders, assessing how well models utilize regions in HRI. We conduct extensive experiments involving 28 VLMs, including Gemini 2.0 Flash and GPT-4o. Experiments on HRScene show that current VLMs achieve an average accuracy of around 50% on real-world tasks, revealing significant gaps in HRI understanding. Results on synthetic datasets reveal that VLMs struggle to effectively utilize HRI regions, showing significant Regional Divergence and lost-in-middle, shedding light on future research. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_18406 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | HRScene: How Far Are VLMs from Effective High-Resolution Image Understanding? Zhang, Yusen Zheng, Wenliang Madasu, Aashrith Shi, Peng Kamoi, Ryo Zhou, Hao Zou, Zhuoyang Zhao, Shu Das, Sarkar Snigdha Sarathi Gupta, Vipul Lu, Xiaoxin Zhang, Nan Zhang, Ranran Haoran Iyer, Avitej Lou, Renze Yin, Wenpeng Zhang, Rui Computation and Language High-resolution image (HRI) understanding aims to process images with a large number of pixels, such as pathological images and agricultural aerial images, both of which can exceed 1 million pixels. Vision Large Language Models (VLMs) can allegedly handle HRIs, however, there is a lack of a comprehensive benchmark for VLMs to evaluate HRI understanding. To address this gap, we introduce HRScene, a novel unified benchmark for HRI understanding with rich scenes. HRScene incorporates 25 real-world datasets and 2 synthetic diagnostic datasets with resolutions ranging from 1,024 $\times$ 1,024 to 35,503 $\times$ 26,627. HRScene is collected and re-annotated by 10 graduate-level annotators, covering 25 scenarios, ranging from microscopic to radiology images, street views, long-range pictures, and telescope images. It includes HRIs of real-world objects, scanned documents, and composite multi-image. The two diagnostic evaluation datasets are synthesized by combining the target image with the gold answer and distracting images in different orders, assessing how well models utilize regions in HRI. We conduct extensive experiments involving 28 VLMs, including Gemini 2.0 Flash and GPT-4o. Experiments on HRScene show that current VLMs achieve an average accuracy of around 50% on real-world tasks, revealing significant gaps in HRI understanding. Results on synthetic datasets reveal that VLMs struggle to effectively utilize HRI regions, showing significant Regional Divergence and lost-in-middle, shedding light on future research. |
| title | HRScene: How Far Are VLMs from Effective High-Resolution Image Understanding? |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2504.18406 |