Released: Mar 31, 2026

Atualizações na V26.3

Correções

• Performance and memory consumption issue: add 'EnableCpuMemArena' and 'EnableMemoryPattern' to configure ONNX Runtime.

Released: Feb 27, 2026

Atualizações na V26.2

Recursos

• Internal improvements and stability refinements.

Released: Feb 1, 2026

Atualizações na V26.1

Recursos

• Added RecognizeTables method in the API.

Released: Nov 19, 2025

Atualizações na V25.11

Recursos

• Added CSV output format support for table detection (in OcrOutput class).
• Added support for OCRTable recognition result format.
• Integrated formula detection and recognition functionality.

Correções

• Fixed bug with PDF recognition in the Docker based on the customer's issue.

Released: Oct 9, 2025

Atualizações na V25.10

Recursos

• Enhanced setUpscaleSmallFont behavior in RecognitionSettings to improve recognition quality.
• Updated Aspose.PDF dependency to version 25.9 to improve stability and compatibility.

Released: Sep 17, 2025

Atualizações na V25.9

Recursos

• Enhanced Markdown output and layout detection, resulting in more accurate table and image extraction and improved formatting.

Released: Aug 31, 2025

Atualizações na V25.8

Recursos

• Enhanced Excel output and table layout detection when using DetectAreasMode.TABLE, resulting in more accurate structure extraction and improved formatting.
• Updated ONNX Runtime to version 1.22.1 for improved performance, stability, and compatibility.

Released: Jun 16, 2025

Atualizações na V25.6

Recursos

• Added a debug mode to the API to allow users to view detection area results on the image.
• Added 'confidence' setting for text lines in RecognitionResult.

Released: May 26, 2025

Atualizações na V25.5

Recursos

• Exposed control over ONNX session options for advanced users.
• Optimization: Created classes for one-time InferenceSession initialization and optimized performance.

Released: Apr 15, 2025

Atualizações na V25.4

Recursos

• Added universal recognition of Arabic, Persian and English alphabets.
• Added support for the automatic analysis of image content and detection of layout blocks.
• Added support for markdown output format with document layout.
• Automatic language detection during recognition.
• Improved DOCX output format.