List your company
ResearchFormerly WizWand

Scene Text Visual Question Answering on ST-VQA

68.96AccuracyQwen 2.5 VL + ViCrop (rel-att)
51.820856.270460.7265.1696Nov 25, 2025
Updated 7mo ago

Evaluation results

MethodLinks
Qwen 2.5 VL + ViCrop (rel-att)Base Model=Qwen 2.5 VL, Cropping Strategy=ViCrop (rel-att), Input Resolution=448x4482025.1168.96
Qwen 2.5 VL + CropVLMBase Model=Qwen 2.5 VL, Cropping Strategy=CropVLM, Input Resolution=2048x20482025.1168.31
Qwen 2.5 VL + ViCrop (grad-att)Base Model=Qwen 2.5 VL, Cropping Strategy=ViCrop (grad-att), Input Resolution=448x4482025.1168.09
Qwen 2.5 VL + UV-CoTBase Model=Qwen 2.5 VL, Cropping Strategy=UV-CoT, Input Resolution=336x3362025.1167.91
Qwen 2.5 VLBase Model=Qwen 2.5 VL, Cropping Strategy=None, Input Resolution=448x4482025.1165.49
LLaVA 1.5 + UV-CoTBase Model=LLaVA 1.5, Cropping Strategy=UV-CoT, Input Resolution=336x3362025.1159.3
LLaVA 1.5 + ViCrop (grad-att)Base Model=LLaVA 1.5, Cropping Strategy=ViCrop (grad-att), Input Resolution=336x3362025.1157.06
LLaVA 1.5 + ViCrop (rel-att)Base Model=LLaVA 1.5, Cropping Strategy=ViCrop (rel-att), Input Resolution=336x3362025.1156.95
LLaVA 1.5 + CropVLMBase Model=LLaVA 1.5, Cropping Strategy=CropVLM, Input Resolution=2048x20482025.1156.81
LLaVA 1.5Base Model=LLaVA 1.5, Cropping Strategy=None, Input Resolution=336x3362025.1152.48