1. Products
  2.   Aspose.OCR
  3.   Aspose.OCR FOSS for .NET

Aspose.OCR FOSS for .NET

Open-source .NET library for detecting and recognizing text in images and PDF documents. MIT licensed, runs offline with embedded ONNX models.

Aspose.OCR FOSS for .NET Aspose.OCR FOSS
for .NET

Documentation GitHub

Open-Source .NET Library for Text Detection and Recognition

Aspose.OCR FOSS for .NET is a free, MIT-licensed C# library for detecting and recognizing text in raster images and PDF documents, without requiring external OCR software or services. The entire pipeline — text-line detection, region refinement, and Latin-script recognition — runs locally through three ONNX models embedded directly in the assembly and executed on the CPU via Microsoft.ML.OnnxRuntime; no model download and no network call are needed at runtime. The API is centered on AsposeOcr.Recognize(), which accepts an OcrInput built from one or more files (InputType.SingleImage or InputType.PDF) and returns an OcrOutput — one RecognitionResult per image or PDF page, in source order. Each RecognitionResult exposes the complete page text through RecognitionText and the individual recognized lines, each with its pixel-coordinate bounding box, through RecognitionLinesResult. PDF pages are rendered internally (via Aspose.PDF.FOSS) before recognition, so no separate PDF-to-image conversion step is needed. Aspose.OCR FOSS for .NET is MIT licensed. Install it with dotnet add package Aspose.OCR.FOSS. For the enterprise product family, see Aspose.OCR — Enterprise Product Family.

Recognize Text in Images and PDFs

  • AsposeOcr.Recognize(OcrInput) is the single entry point for both images and PDFs
  • OcrInput accepts InputType.SingleImage or InputType.PDF; add one or more file paths with OcrInput.Add()
  • For InputType.PDF, every page is recognized and returned in source order
  • The result is an OcrOutput — one RecognitionResult per image or PDF page

Where Text Recognition Is Used

  • Digitizing scanned documents into searchable, copyable text
  • Extracting text from scanned PDFs that have no embedded text layer
  • Processing scanned forms, receipts, and invoices in a batch pipeline
  • Indexing large archives of scanned images for full-text search

Line-Level Text and Bounding Boxes

  • RecognitionResult.RecognitionText returns the full recognized page text
  • RecognitionResult.RecognitionLinesResult returns one RecognitionResult.LinesResult per detected text line
  • Each RecognitionResult.LinesResult.TextInLine holds that line’s recognized text
  • Each RecognitionResult.LinesResult.Line is a System.Drawing.Rectangle giving the line’s pixel coordinates in the source image

Where Line-Level Data Is Used

  • Drawing recognized-region overlays on top of the source image
  • Building a searchable text layer positioned over a scanned page
  • Reviewing per-line recognition accuracy during QA
  • Driving redaction or annotation tooling from line coordinates

Runs Fully Offline, No External Service

  • All three ONNX models (detection, refinement, Latin recognition) are embedded assembly resources — nothing to download separately
  • Recognition runs entirely on the local CPU through Microsoft.ML.OnnxRuntime
  • OcrInput can hold multiple files; OcrOutput preserves their original order
  • Supported input types are InputType.SingleImage and InputType.PDF only — any other input type throws NotSupportedException

Where Offline Recognition Matters

  • Air-gapped or regulated environments where documents cannot leave the network
  • CI/CD pipelines without external network access during the build
  • High-volume batch recognition where per-call cloud OCR pricing is a concern
  • On-premises document processing with predictable, local-only latency

Recognize Text in a PDF Document

Recognize every page of a PDF and print each page’s text in source order.

using Aspose.OCR;
var api = new AsposeOcr();
using var input = new OcrInput(InputType.PDF);
input.Add("document.pdf");
OcrOutput output = api.Recognize(input);
foreach (RecognitionResult page in output)
{
    Console.WriteLine(page.RecognitionText);
}

Recognize Text in a Single Image

Recognize a single raster image and print its recognized text.

using Aspose.OCR;
var api = new AsposeOcr();
using var input = new OcrInput(InputType.SingleImage);
input.Add("scan.png");
RecognitionResult result = api.Recognize(input)[0];
Console.WriteLine(result.RecognitionText);

Read Recognized Lines and Their Bounding Boxes

Iterate the recognized lines and print each line’s text with its pixel coordinates.

using Aspose.OCR;
var api = new AsposeOcr();
using var input = new OcrInput(InputType.SingleImage);
input.Add("document.jpg");
RecognitionResult result = api.Recognize(input)[0];
foreach (RecognitionResult.LinesResult line in result.RecognitionLinesResult)
{
    System.Drawing.Rectangle bounds = line.Line;
    Console.WriteLine($"x={bounds.X}, y={bounds.Y}, width={bounds.Width}, height={bounds.Height}: {line.TextInLine}");
}

Frequently Asked Questions

What is the licensing model?

Aspose.OCR FOSS for .NET is MIT licensed. You can use, modify, and redistribute it, including in commercial products, without additional licensing fees.

How do I install Aspose.OCR FOSS for .NET?

See the install section above for the current install command.

What languages does it recognize?

Latin-script text. There is no language-selection parameter, and non-Latin scripts are not covered by the embedded recognition model.

Does it require an internet connection?

No. All three ONNX models (detection, refinement, recognition) are embedded in the assembly and run locally through Microsoft.ML.OnnxRuntime.

Can I recognize multiple files in one call?

Yes — add multiple files to the same OcrInput before calling AsposeOcr.Recognize(). The returned OcrOutput preserves their order.

  

Support and Learning Resources