ITextDetector<T>

Interfaces · AiDotNet.Document.Interfaces

Interface for text detection models that locate text regions in images.

Implements: IDocumentModel<T>

Known implementations: PSENet<T>, TRIE<T>

For Beginners

Text detection is the first step in reading text from images. It’s like highlighting all the places where text appears, but not actually reading it. After detection, a text recognizer reads the actual characters in each highlighted region.

Example usage:

How It Works

Text detection models find where text appears in an image without reading the text. They output bounding boxes (polygons or rectangles) around text regions.

Properties

Property Summary
bool SupportsPolygonOutput { get; } Gets whether this detector outputs polygon bounding boxes (vs axis-aligned rectangles).
bool SupportsRotatedText { get; } Gets whether this detector supports rotated text detection.
int MinTextHeight { get; } Gets the minimum detectable text height in pixels.

Methods

Method Summary
Tensor<T> GetProbabilityMap(Tensor<T> image) Gets the probability map showing text likelihood at each pixel.
TextDetectionResult<T> DetectText(Tensor<T> image) Detects text regions in an image.
TextDetectionResult<T> DetectText(Tensor<T> image, double confidenceThreshold) Detects text regions with a custom confidence threshold.