FireRed-OCR Technical Report

2026-03-02

AI Summary

This work addresses the structural hallucination issues of general-purpose vision-language models (VLMs) in complex document parsing, which hinder their ability to meet industrial OCR demands for pixel-level accuracy. To this end, we propose FireRed-OCR, a framework built upon Qwen3-VL that transforms a general VLM into a high-precision expert for structured document understanding. Our approach features a “geometry + semantics” data factory and a three-stage progressive training strategy: multi-task pre-alignment, instruction fine-tuning, and format-constrained Group Relative Policy Optimization. We innovatively integrate geometric feature clustering with multidimensional labeling to synthesize high-quality training data and introduce a format-aware reinforcement learning mechanism to enforce strict compliance with output structure and syntax. Evaluated on OmniDocBench v1.5, FireRed-OCR achieves a state-of-the-art overall score of 92.94%, outperforming DeepSeek-OCR 2 and OCRVerse across text, formula, table, and reading order metrics.

Technology Category

• Computer Vision: Large Vision Models
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

We present FireRed-OCR, a systematic framework to specialize general VLMs into high-performance OCR models. Large Vision-Language Models (VLMs) have demonstrated impressive general capabilities but frequently suffer from ``structural hallucination'' when processing complex documents, limiting their utility in industrial OCR applications. In this paper, we introduce FireRed-OCR, a novel framework designed to transform general-purpose VLMs (based on Qwen3-VL) into pixel-precise structural document parsing experts. To address the scarcity of high-quality structured data, we construct a ``Geometry + Semantics'' Data Factory. Unlike traditional random sampling, our pipeline leverages geometric feature clustering and multi-dimensional tagging to synthesize and curate a highly balanced dataset, effectively handling long-tail layouts and rare document types. Furthermore, we propose a Three-Stage Progressive Training strategy that guides the model from pixel-level perception to logical structure generation. This curriculum includes: (1) Multi-task Pre-alignment to ground the model's understanding of document structure; (2) Specialized SFT for standardizing full-image Markdown output; and (3) Format-Constrained Group Relative Policy Optimization (GRPO), which utilizes reinforcement learning to enforce strict syntactic validity and structural integrity (e.g., table closure, formula syntax). Extensive evaluations on OmniDocBench v1.5 demonstrate that FireRed-OCR achieves state-of-the-art performance with an overall score of 92.94\%, significantly outperforming strong baselines such as DeepSeek-OCR 2 and OCRVerse across text, formula, table, and reading order metrics. We open-source our code and model weights to facilitate the ``General VLM to Specialized Structural Expert'' paradigm.

Problem

• structural hallucination
• OCR
• Vision-Language Models
• document parsing
• industrial OCR

Innovation

• Geometry + Semantics Data Factory
• Three-Stage Progressive Training
• Format-Constrained GRPO
• Structural Document Parsing
• VLM Specialization