ITextRecognizer<T>

Interfaces · AiDotNet.Document.Interfaces

Interface for text recognition models that read text from cropped image regions.

Implements: IDocumentModel<T>

Known implementations: ABINet<T>, SVTR<T>

For Beginners

Text recognition is the second step in reading text from images. Given a small image containing only text (like a single word or line), the recognizer outputs the actual characters. This is like reading what’s written in a highlighted region.

Example usage:

How It Works

Text recognition models convert cropped images of text into character sequences. They work on pre-detected text regions (from a text detector).

Properties

Property Summary
bool SupportsAttentionVisualization { get; } Gets whether this recognizer supports attention visualization.
int MaxSequenceLength { get; } Gets the maximum sequence length this recognizer can output.
string SupportedCharacters { get; } Gets the supported character set (alphabet) for this recognizer.

Methods

Method Summary
IEnumerable<TextRecognitionResult<T>> RecognizeTextBatch(IEnumerable<Tensor<T>> croppedImages) Recognizes text from multiple cropped image regions (batch processing).
Tensor<T> GetAttentionWeights() Gets the attention weights for visualization (if supported).
Tensor<T> GetCharacterProbabilities() Gets the character-level probabilities for the last recognition.
TextRecognitionResult<T> RecognizeText(Tensor<T> croppedImage) Recognizes text from a cropped image region.