

Marketplace

Pre-Installed PaddleOCR

Get PaddleOCR pre-installed and ready to go from the Database Mart Marketplace—just deploy and use, no setup hassle.

Overview

PaddleOCR is an industry-leading, open-source optical character recognition (OCR) and document AI toolkit developed by the PaddlePaddle team at Baidu. It includes a comprehensive range of models—from the ultra-lightweight PP-OCR series for efficient text recognition, to the 0.9B-parameter PaddleOCR-VL vision-language model for advanced document parsing supporting 109 languages, and the PP-Structure pipeline for complex PDF and image parsing into structured JSON and Markdown formats.

Select a pre-installed PaddleOCR Marketplace image and get a fully configured server that's ready to use immediately—no manual installation or complex setup required.

Supported Operating Systems

The PaddleOCR Marketplace image is available for Ubuntu Server 22 LTS 64-bit.

Supported Products / Plans

The PaddleOCR Marketplace image is currently available across the following plans:

GPU VPS

Professional GPU VPS - RTX A4000
Advanced GPU VPS - RTX 5090

GPU Dedicated Server

Basic Dedicated GPU Server - RTX 5060
Advanced Dedicated GPU Server - RTX 3060 Ti
Advanced Dedicated GPU Server - RTX A5000
Enterprise Dedicated GPU Server - RTX 5090
Enterprise Dedicated GPU Server - RTX 4090

View compatible plans →

PaddleOCR Version

Database Mart provides the v2.7 version of PaddleOCR, pre-installed and ready to use.

What's Installed

When you select the PaddleOCR Marketplace image, Database Mart automatically installs and configures:

Core Operating Environment

Operating System: Ubuntu Server 22.04 LTS, pre-optimized for OCR and document AI workloads
GPU Drivers & Acceleration: NVIDIA drivers, CUDA 11.2+, and cuDNN pre-installed for GPU acceleration
Python Environment: Python 3.8 with a dedicated virtual environment (Conda or venv) for dependency management

PaddleOCR Core Components

PaddlePaddle Framework: PaddlePaddle 3.2.0+ (GPU version) pre-installed with CUDA support
PaddleOCR Toolkit: PaddleOCR 2.7 (or newer) installed via pip with full functionality
Pre-trained Models: PP-OCRv5 detection, recognition, and direction classification models pre-downloaded and configured for immediate use
Model Paths: Models stored in ./inference/ directory with pre-configured paths for detection (ch_PP-OCRv5_det_infer), recognition (ch_PP-OCRv5_rec_infer), and classification models

Installation Process

During server provisioning, Database Mart automatically performs the following steps:

Deploys the selected Linux operating system (Ubuntu Server 22.04 LTS).
Installs system dependencies, including build-essential, python3-dev, OpenCV, and other essential libraries.
Creates a Python virtual environment (Conda or venv) and installs core dependencies:
- PaddlePaddle-GPU 3.2.0+ with CUDA 11.2+ support
- PaddleOCR toolkit (full installation with paddleocr[all])
- Additional libraries including OpenCV, NumPy, and Flask/FastAPI for API deployment
Downloads and configures the selected pre-trained models:
- PP-OCRv5 detection, recognition, and direction classification models
- Model files are stored in ./inference/ with pre-configured paths

Everything is fully automated—your PaddleOCR server is ready to use as soon as deployment is complete.

Getting Started After Deployment

After your server has been provisioned, follow these steps to begin using PaddleOCR.

Step 1: Access the Server

Once deployment is complete, connect to your server via SSH using the credentials provided by Database Mart:

ssh administrator@your-server-ip

Step 2: Activate the Environment and Verify Installation

# Activate the pre-configured virtual environment
cd /opt/paddleocr
source venv/bin/activate

# Verify the installation
from paddleocr import PaddleOCR
ocr = PaddleOCR(use_angle_cls=True, lang="ch")
print("PaddleOCR installed successfully!")
Expected output: Confirmation that the installation is complete and ready for use.

Step 3: Test Basic Text Recognition

Create a test Python script to verify OCR functionality:

from paddleocr import PaddleOCR

# Initialize OCR with Chinese model and angle classification
ocr = PaddleOCR(use_angle_cls=True, lang="ch")  # lang="ch" for Chinese, "en" for English [citation:2]

# Test with an image
img_path = "test.jpg"  # Replace with your image path
result = ocr.ocr(img_path, cls=True)

# Print results
for line in result:
    print(f"Coordinates: {line[0]}")
    print(f"Text: {line[1][0]}")
    print(f"Confidence: {line[1][1]:.2f}")

Step 4: Use Advanced Document Parsing

For PDF or complex document processing, use the PP-Structure pipeline:

# Parse a PDF document into Markdown format
paddlex --pipeline OCR --device gpu --input document.pdf --output ./output/

Or using Python:

from paddleocr import PaddleOCR

# Initialize with structure parsing enabled
ocr = PaddleOCR(use_angle_cls=True, lang="ch", enable_structure=True)

# Process a PDF or image with layout parsing
result = ocr.ocr("document.pdf", cls=True, structure=True)

# Results include text, tables, formulas, and layout structure

Step 5: Start the REST API Server

The server comes with pre-configured service deployment options:
Using PaddleX (Recommended for PaddleOCR 3.x):

# Start the OCR service
paddlex --serve --pipeline OCR --device gpu --host 0.0.0.0 --port 8080

The API server will be accessible at http://your-server-ip:8080.

Using PaddleHub Serving (For PaddleOCR 2.x):

# Install the OCR system module
hub install deploy/hubserving/ocr_system

# Start the service
hub serving start -m ocr_system --port 8866

Step 6: Send a Prediction Request

Once the API server is running, test it with a request:

python tools/test_hubserving.py \
  --server_url=http://your-server-ip:8080/predict/ocr_system \
  --image_dir=./doc/imgs/

Or using curl:

curl -X POST "http://localhost:8080/predict/ocr_system" \
  -F "image=@test.jpg" \
  -F "use_gpu=true" \
  --output result.json

Notes

GPU Acceleration: Ensure the CUDA environment is correctly configured. Use CUDA_VISIBLE_DEVICES to specify GPU cards.
Batch Processing: Increase the rec_batch_num parameter to improve GPU utilization for batch inference.
Model Quantization: Use PaddleSlim to convert FP32 models to INT8 for faster inference.
Memory Management: Reduce rec_batch_num if encountering out-of-memory errors.

Frequently Asked Questions

Q: What PaddleOCR versions are available?

Database Mart provides PaddleOCR 2.7 with pre-configured PP-OCRv5 models.

Q: What languages does PaddleOCR support?

PaddleOCR supports 109 languages, including Chinese, English, Japanese, Korean, French, German, Arabic, Hindi, Russian, and more. Multilingual support is built into PP-OCRv5 models.