Spire.OCR 2.3.1 focuses on improvements for the .NET Standard platform. This release adds support for PaddleOCR 6.0.0 models, which no longer require spire_ocrsystem.dll and are available in three sizes: Tiny, Small, and Medium. The new models also support cross-platform deployment without requiring separate model configurations for different operating systems.

Summary of Updates

New Feature

Support for PaddleOCR 6.0.0 Models

Spire.OCR 2.3.1 supports PaddleOCR 6.0.0 ONNX models. Unlike the previous model configuration, the new models do not depend on spire_ocrsystem.dll.

Three model sizes are available to accommodate different deployment requirements:

• Tiny: For lightweight deployments.
• Small: For general-purpose use.
• Medium: For deployments that require a larger model.

The models support cross-platform use without requiring operating-system-specific model variants.

Model downloads:

• Direct download: PaddleOCR 6.0.0 ONNX Models
• Hugging Face: pp-ocrv6-onnx

Code Example

The following example configures the OCR scanner to use the Small model, sets English as the recognition language, extracts text from an image, and saves the result to a text file.

using System.IO;
using Spire.OCR;

OcrScanner scanner = new OcrScanner();

// Configure the scanner.
ConfigureOptions configureOptions = new ConfigureOptions();

// Set the path to the PaddleOCR 6.0.0 model.
configureOptions.ModelPath = @"Models\6.0.0\small";

// Set the recognition language.
configureOptions.Language = "English";

// Apply the configuration.
scanner.ConfigureDependencies(configureOptions);

// Recognize text in the image.
scanner.Scan("in.png");

// Save the recognized text to a file.
string text = scanner.Text.ToString();
File.WriteAllText("Output.txt", text);

Dependency Updates for .NET Standard

This release adjusts the dependencies required for .NET Standard projects. The following package versions are specified as minimum requirements: