PP-OCRv6: From 1.5M to 34.5M Parameters, Surpassing Billion-Scale VLMs on OCR Tasks

2026-06-11

AI Summary

This work addresses prevalent challenges in current vision-language models for OCR tasks—namely hallucination, inaccurate localization, and high computational overhead—by introducing a lightweight OCR system. Built upon the MetaFormer architecture, the proposed approach features a unified structural reparameterization module that decouples spatial and channel mixing and supports multi-scale task configurations. Coupled with task-specific strides and data-centric optimization strategies, the system enables seamless deployment across diverse scenarios, from servers to edge devices. The medium-sized model achieves an 83.2% recognition accuracy and an 86.2% detection H-mean, substantially outperforming prior methods and even billion-parameter vision-language models. Meanwhile, the tiny variant delivers a 3.9× speedup on CPU inference with comparable accuracy.

Abstract

Vision-Language Models (VLMs) have achieved impressive results on general vision-language tasks, yet they suffer from hallucination, imprecise localization, and prohibitive computational cost when applied to dedicated OCR scenarios. This paper presents PP-OCRv6, a lightweight OCR system that combines architectural innovation with data-centric optimization. PP-OCRv6 redesigns the backbone, detection neck, and recognition neck around a unified MetaFormer-style building block with structural reparameterization, decoupling spatial token mixing from channel mixing and supporting both tasks through task-specific stride configurations. Three model tiers (medium, small, tiny) share the same block primitives, covering deployment scenarios from server to edge. On our in-house benchmarks, PP-OCRv6_medium achieves 83.2% recognition accuracy and 86.2% detection Hmean, outperforming PP-OCRv5_server by +5.1% and +4.6% respectively while surpassing Qwen3-VL-235B, GPT-5.5, and Gemini-3.1-Pro with orders of magnitude fewer parameters. The tiny tier achieves 3.9$\times$ faster inference than PP-OCRv5_mobile on Intel Xeon CPU while maintaining comparable accuracy.

Problem

• OCR
• Vision-Language Models
• hallucination
• localization
• computational cost

Innovation

• MetaFormer
• structural reparameterization
• lightweight OCR
• task-specific stride
• vision-language models