# Scene Text Visual Question Answering on ST-VQA Source: [Scene Text Visual Question Answering on ST-VQA]() Collection index: [llms.txt]() Data updated: 2026-04-15T07:32:53.231Z Evaluation of model performance on the ST-VQA dataset for Scene Text Visual Question Answering, measured by Accuracy. 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. | Scope | Recorded value | | --- | --- | | Task | Scene Text Visual Question Answering | | Dataset (including any stated split) | ST-VQA | | Metric | Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 10 of 10 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Accuracy](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy | Result date | Report paper | | --- | --- | --- | --- | | Qwen 2.5 VL + ViCrop (rel-att) | 68.96 | Not recorded | 69268982d6b0dd3342930319 | | Qwen 2.5 VL + CropVLM | 68.31 | Not recorded | 69268982d6b0dd3342930319 | | Qwen 2.5 VL + ViCrop (grad-att) | 68.09 | Not recorded | 69268982d6b0dd3342930319 | | Qwen 2.5 VL + UV-CoT | 67.91 | Not recorded | 69268982d6b0dd3342930319 | | Qwen 2.5 VL | 65.49 | Not recorded | 69268982d6b0dd3342930319 | | LLaVA 1.5 + UV-CoT | 59.3 | Not recorded | 69268982d6b0dd3342930319 | | LLaVA 1.5 + ViCrop (grad-att) | 57.06 | Not recorded | 69268982d6b0dd3342930319 | | LLaVA 1.5 + ViCrop (rel-att) | 56.95 | Not recorded | 69268982d6b0dd3342930319 | | LLaVA 1.5 + CropVLM | 56.81 | Not recorded | 69268982d6b0dd3342930319 | | LLaVA 1.5 | 52.48 | Not recorded | 69268982d6b0dd3342930319 | ### Qwen 2.5 VL + ViCrop (rel-att) Accuracy: 68.96. Result ID: 69df3f1647f34ce924c7adec. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen 2.5 VL","Cropping Strategy":"ViCrop (rel-att)","Input Resolution":"448x448"}. This record does not assert independent reproduction. ### Qwen 2.5 VL + CropVLM Accuracy: 68.31. Result ID: 69df3f1647f34ce924c7adef. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen 2.5 VL","Cropping Strategy":"CropVLM","Input Resolution":"2048x2048"}. This record does not assert independent reproduction. ### Qwen 2.5 VL + ViCrop (grad-att) Accuracy: 68.09. Result ID: 69df3f1647f34ce924c7aded. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen 2.5 VL","Cropping Strategy":"ViCrop (grad-att)","Input Resolution":"448x448"}. This record does not assert independent reproduction. ### Qwen 2.5 VL + UV-CoT Accuracy: 67.91. Result ID: 69df3f1647f34ce924c7adee. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen 2.5 VL","Cropping Strategy":"UV-CoT","Input Resolution":"336x336"}. This record does not assert independent reproduction. ### Qwen 2.5 VL Accuracy: 65.49. Result ID: 69df3f1647f34ce924c7adeb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen 2.5 VL","Cropping Strategy":"None","Input Resolution":"448x448"}. This record does not assert independent reproduction. ### LLaVA 1.5 + UV-CoT Accuracy: 59.3. Result ID: 69df3f1647f34ce924c7ade9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"LLaVA 1.5","Cropping Strategy":"UV-CoT","Input Resolution":"336x336"}. This record does not assert independent reproduction. ### LLaVA 1.5 + ViCrop (grad-att) Accuracy: 57.06. Result ID: 69df3f1647f34ce924c7ade8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"LLaVA 1.5","Cropping Strategy":"ViCrop (grad-att)","Input Resolution":"336x336"}. This record does not assert independent reproduction. ### LLaVA 1.5 + ViCrop (rel-att) Accuracy: 56.95. Result ID: 69df3f1647f34ce924c7ade7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"LLaVA 1.5","Cropping Strategy":"ViCrop (rel-att)","Input Resolution":"336x336"}. This record does not assert independent reproduction. ### LLaVA 1.5 + CropVLM Accuracy: 56.81. Result ID: 69df3f1647f34ce924c7adea. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"LLaVA 1.5","Cropping Strategy":"CropVLM","Input Resolution":"2048x2048"}. This record does not assert independent reproduction. ### LLaVA 1.5 Accuracy: 52.48. Result ID: 69df3f1647f34ce924c7ade6. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Base Model":"LLaVA 1.5","Cropping Strategy":"None","Input Resolution":"336x336"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.