Interface ExtractedImage

An image extracted from a document during ingestion.

interface ExtractedImage {
    data: Buffer<ArrayBufferLike>;
    mimeType: string;
    caption?: string;
    pageNumber?: number;
    embedding?: number[];
}

Properties

data mimeType caption? pageNumber? embedding?

Properties

data

data: Buffer<ArrayBufferLike>

Raw image bytes (PNG, JPEG, WebP, etc.).

mimeType

mimeType: string

MIME type of data.

Example

'image/png' | 'image/jpeg'

Optional caption

caption?: string

Auto-generated or OCR-derived caption. Present when a vision LLM is configured and extractImages: true.

Optional pageNumber

pageNumber?: number

Page number the image appears on (1-based, PDF/DOCX).

Optional embedding

embedding?: number[]

Dense embedding of the image caption or visual content. Only present when embeddings were computed during extraction.

Settings

Member Visibility

Theme

On This Page

datamimeTypecaptionpageNumberembedding