PP-OCRv5 Online Demo

PP-OCRv5 is the latest generation of the PP-OCR series model, designed to handle a wide range of scene and text types.
It supports five major text types: Simplified Chinese, Traditional Chinese, Pinyin annotation, English, and Japanese.
PP-OCRv5 has enhanced recognition capabilities for challenging use cases, including complex handwritten Chinese and English, vertical text, and rare characters.
To use it, simply upload your image, or click one of the examples to load them. Read more at the links below.

Since our inference server is deployed in mainland China, cross-border network transmission may be slow, which could result in a suboptimal experience on Hugging Face. We recommend visiting the PaddlePaddle AI Studio Community to try the demo for a smoother experience.

Examples

Advance Options

Using the document image orientation classification module

yes no

Using the document unwarping module

yes no

Using the text line orientation classification module

yes no

Results

🚀 User Guide
Step 1: Upload Your File
Supported formats: JPG, PNG, PDF, JPEG
Step 2: Click Analyze Document Button
System will process automatically
Step 3: Wait for Results
Results will be displayed after processing
Step 4: Download results zip
Results zip will be displayed after processing

Attention: Only the first 10 pages will be processed
⚠️Special Attention: For safety reasons, please make sure that the uploaded files do not contain personal information.

PP-OCRv5 Online Demo

Using the document image orientation classification module

Using the document unwarping module

Using the text line orientation classification module

Image side length restriction type for text detection

Layout region detection expansion coefficient

Text detection pixel threshold

Text detection box threshold

Text detection unclip ratio

Text recognition score threshold

Results