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

TL;DR

PaddleOCR-VL-1.5 is a compact multi-task vision-language model achieving state-of-the-art accuracy on document parsing benchmarks and demonstrating robustness against real-world distortions.

Contribution

The paper introduces PaddleOCR-VL-1.5 with new benchmarks and extended capabilities, maintaining high efficiency and compactness.

Findings

Achieves 94.5% accuracy on OmniDocBench v1.5

Sets new SOTA performance on Real5-OmniDocBench benchmark

Includes seal recognition and text spotting tasks

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