PaddleOCR-VL-1.5: Towards a Multi-Task 0.9B VLM for Robust In-the-Wild Document Parsing

2026-01-29

AI Summary

This work addresses the challenge of insufficient robustness in document image parsing under real-world physical distortions—such as scanning artifacts, perspective skew, screen capture effects, and illumination variations—by proposing a lightweight 0.9B-parameter multi-task vision-language model (VLM). For the first time, this compact architecture jointly integrates document parsing, seal detection, and text detection within a unified framework. To evaluate model robustness under realistic distortions, we introduce the Real5-OmniDocBench benchmark. Our model achieves a state-of-the-art accuracy of 94.5% on OmniDocBench v1.5 and demonstrates leading performance on the new benchmark, significantly enhancing end-to-end document understanding in complex, real-world scenarios.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Safety and Robustness

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

We introduce PaddleOCR-VL-1.5, an upgraded model achieving a new state-of-the-art (SOTA) accuracy of 94.5% on OmniDocBench v1.5. To rigorously evaluate robustness against real-world physical distortions, including scanning, skew, warping, screen-photography, and illumination, we propose the Real5-OmniDocBench benchmark. Experimental results demonstrate that this enhanced model attains SOTA performance on the newly curated benchmark. Furthermore, we extend the model's capabilities by incorporating seal recognition and text spotting tasks, while remaining a 0.9B ultra-compact VLM with high efficiency. Code: https://github.com/PaddlePaddle/PaddleOCR

Problem

• document parsing
• robustness
• real-world distortions
• multi-task learning
• visual language model

Innovation

• multi-task VLM
• robust document parsing
• Real5-OmniDocBench
• seal recognition
• ultra-compact model