OmniDocBench v-1.5

OmniDocBench v1.5 is a benchmark dataset used to evaluate the robustness of document parsing models against real-world physical distortions such as scanning, skew, warping, screen-photography, and illumination.

Papers using OmniDocBench v-1.5 (3)

PaddleOCR-VL-1.5: Towards a Multi-Task 0.9B VLM for Robust In-the-Wild Document Parsing2026FireRed-OCR Technical Report2026Qianfan-OCR: A Unified End-to-End Model for Document Intelligence2026