Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2602.04802 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916017277501440 |
|---|---|
| author | Liu, Qing'an Feng, Juntong Wang, Yuhao Han, Xinzhe Cheng, Yujie Zhu, Yue Diao, Haiwen Zhuge, Yunzhi Lu, Huchuan |
| author_facet | Liu, Qing'an Feng, Juntong Wang, Yuhao Han, Xinzhe Cheng, Yujie Zhu, Yue Diao, Haiwen Zhuge, Yunzhi Lu, Huchuan |
| contents | Vision-Language Models (VLMs) have achieved impressive performance in cross-modal understanding across textual and visual inputs, yet existing benchmarks predominantly focus on pure-text queries. In real-world scenarios, language also frequently appears as visualized text embedded in images, raising the question of whether current VLMs handle such input requests comparably. We introduce VISTA-Bench, a systematic benchmark from multimodal perception, reasoning, to unimodal understanding domains. It evaluates visualized text understanding by contrasting pure-text and visualized-text questions under controlled rendering conditions. Extensive evaluation of over 30 representative VLMs reveals a pronounced modality gap: models that perform well on pure-text queries often degrade substantially when equivalent semantic content is presented as visualized text. This gap is further amplified by increased perceptual difficulty, highlighting sensitivity to rendering variations despite unchanged semantics. Overall, VISTA-Bench provides a principled evaluation framework to diagnose this limitation and to guide progress toward more unified language representations across tokenized text and pixels. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_04802 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | VISTA-Bench: Do Vision-Language Models Really Understand Visualized Text as Well as Pure Text? Liu, Qing'an Feng, Juntong Wang, Yuhao Han, Xinzhe Cheng, Yujie Zhu, Yue Diao, Haiwen Zhuge, Yunzhi Lu, Huchuan Computer Vision and Pattern Recognition Vision-Language Models (VLMs) have achieved impressive performance in cross-modal understanding across textual and visual inputs, yet existing benchmarks predominantly focus on pure-text queries. In real-world scenarios, language also frequently appears as visualized text embedded in images, raising the question of whether current VLMs handle such input requests comparably. We introduce VISTA-Bench, a systematic benchmark from multimodal perception, reasoning, to unimodal understanding domains. It evaluates visualized text understanding by contrasting pure-text and visualized-text questions under controlled rendering conditions. Extensive evaluation of over 30 representative VLMs reveals a pronounced modality gap: models that perform well on pure-text queries often degrade substantially when equivalent semantic content is presented as visualized text. This gap is further amplified by increased perceptual difficulty, highlighting sensitivity to rendering variations despite unchanged semantics. Overall, VISTA-Bench provides a principled evaluation framework to diagnose this limitation and to guide progress toward more unified language representations across tokenized text and pixels. |
| title | VISTA-Bench: Do Vision-Language Models Really Understand Visualized Text as Well as Pure Text? |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2602.04802 |