This repository contains ONNX format models converted from PaddleOCR, a practical ultra-lightweight OCR system. These models are optimized for production deployment and cross-platform compatibility.

📦 Model Files

File Name	Size	Description
`PP-OCRv5_server_det_infer.onnx`	84MB	Text detection model - locates text regions in images
`PP-OCRv5_server_rec_infer.onnx`	81MB	Text recognition model - recognizes text content
`UVDoc_infer.onnx`	30MB	Document rectification model - corrects document perspective
`PP-LCNet_x1_0_doc_ori_infer.onnx`	6.5MB	Document orientation detection
`PP-LCNet_x1_0_textline_ori_infer.onnx`	6.5MB	Text line orientation detection
`PP-OCRv5_server_rec_infer.yml`	145KB	Recognition model configuration file

Total Size: ~208MB

🚀 Quick Start

Installation

pip install huggingface_hub onnxruntime

Download Models

python

from huggingface_hub import hf_hub_download
import os

def download_paddleocr_models():
    """Download all PaddleOCR ONNX models"""
    model_files = [
        "PP-OCRv5_server_det_infer.onnx",
        "PP-OCRv5_server_rec_infer.onnx", 
        "UVDoc_infer.onnx",
        "PP-LCNet_x1_0_doc_ori_infer.onnx",
        "PP-LCNet_x1_0_textline_ori_infer.onnx",
        "PP-OCRv5_server_rec_infer.yml"
    ]
    
    cache_dir = "models"
    os.makedirs(cache_dir, exist_ok=True)
    
    for file in model_files:
        print(f"Downloading {file}...")
        hf_hub_download(
            repo_id="marsena/paddleocr-test",
            filename=file,
            local_dir=cache_dir
        )
    print("All models downloaded!")

# Download models
download_paddleocr_models()

Basic Usage

python

import onnxruntime as ort
import numpy as np
from PIL import Image

# Load detection model
det_session = ort.InferenceSession("models/PP-OCRv5_server_det_infer.onnx")

# Load recognition model  
rec_session = ort.InferenceSession("models/PP-OCRv5_server_rec_infer.onnx")

# Your OCR pipeline implementation here...

🏷️ Model Tags

Framework: ONNX
Task: Computer Vision, OCR
Language: Multi-language support
Domain: Text Detection, Text Recognition

🔧 Technical Details

Conversion Process

These models were converted from PaddlePaddle format to ONNX format for broader compatibility:

Source: Original PaddleOCR models from PaddlePaddle Hub
Conversion: PaddlePaddle → ONNX format
Optimization: Model optimization for inference speed
Validation: Output consistency verification

System Requirements

Runtime: ONNX Runtime
Python: 3.7+
Memory: Minimum 2GB RAM recommended
Platform: Cross-platform (Windows, Linux, macOS)

📄 License

This project follows the Apache 2.0 License, consistent with the original PaddleOCR project.

Original PaddleOCR License

Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

    http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.

🙏 Acknowledgments

Original Project: PaddleOCR by PaddlePaddle Team
Framework: PaddlePaddle
Conversion Tools: ONNX ecosystem

📚 Citation

If you use these models in your research, please cite the original PaddleOCR paper:

bibtex

@misc{paddleocr2020,
    title={PaddleOCR: Awesome multilingual OCR toolkits},
    author={PaddlePaddle Authors},
    year={2020},
    howpublished={\url{https://github.com/PaddlePaddle/PaddleOCR}}
}

❓ Issues & Support

For issues related to:

Model conversion: Create an issue in this repository
Original PaddleOCR: Visit PaddleOCR Issues
ONNX Runtime: Visit ONNX Runtime Issues

Note: This is a community contribution for easier deployment of PaddleOCR models. For production use, please ensure compliance with your specific requirements and test thoroughly.