JinaDocumentImageEmbedder

Compute document embeddings based on images using Jina AI multimodal models. Use this component in indexing pipelines to embed image documents before writing them to a document store.

Key Features

• Uses Jina AI multimodal models (including jina-clip-v2 and jina-embeddings-v4) to embed documents containing images or PDFs.
• Stores the computed embedding in each document's embedding field, enabling image-based semantic search and retrieval.
• Reads image file paths from a configurable document metadata field (default: file_path).
• Supports JPEG and PNG image files, as well as PDF pages converted to images automatically.
• Configurable embedding dimensions and image resizing for efficient storage and processing.
• Supports both synchronous and asynchronous operation via run() and run_async().

Configuration

1. Drag the JinaDocumentImageEmbedder component onto the canvas from the Component Library.
2. Click on the component to open the configuration panel.
3. On the General tab:
   1. Create a secret with your Jina API key. Use JINA_API_KEY as the secret key. For instructions, see Create Secrets. Get your API key from Jina AI.
   2. Set the model field. Supported models include jina-clip-v1, jina-clip-v2 (default), and jina-embeddings-v4. Note that embedding_dimension is only supported by jina-embeddings-v4.
   3. Set the file_path_meta_field to the document metadata key that holds the image file path (default: file_path).
4. Go to the Advanced tab to configure root_path, embedding_dimension, image_size, and batch_size.

Connections

JinaDocumentImageEmbedder receives a list of documents through its documents input. Each document must include an image file path in its metadata under the field specified by file_path_meta_field. It outputs the same documents with their embeddings added through its documents output.

Connect it after a converter that creates documents with image paths in their metadata. Connect its documents output to DocumentWriter to store embedded documents in a document store.

Source Code

To check this component's source code, open document_image_embedder.py in the Haystack Core Integrations repository.

Usage Examples

Basic Configuration

  JinaDocumentImageEmbedder:
    type: haystack_integrations.components.embedders.jina.document_image_embedder.JinaDocumentImageEmbedder
    init_parameters:
      api_key:
        type: env_var
        env_vars:
        - JINA_API_KEY
        strict: false
      model: jina-clip-v2
      file_path_meta_field: file_path
      batch_size: 5

Using the Component in a Pipeline

This example shows an indexing pipeline that takes documents with image file paths in their metadata, embeds the images using Jina, and writes them to a document store.

# haystack-pipeline
components:
  JinaDocumentImageEmbedder:
    type: haystack_integrations.components.embedders.jina.document_image_embedder.JinaDocumentImageEmbedder
    init_parameters:
      api_key:
        type: env_var
        env_vars:
        - JINA_API_KEY
        strict: false
      model: jina-clip-v2
      file_path_meta_field: file_path
      root_path:
      embedding_dimension:
      image_size:
      batch_size: 5
  DocumentWriter:
    type: haystack.components.writers.document_writer.DocumentWriter
    init_parameters:
      policy: OVERWRITE
      document_store:
        type: haystack_integrations.document_stores.opensearch.document_store.OpenSearchDocumentStore
        init_parameters:
          hosts:
          index: Image-Index
          max_chunk_bytes: 104857600
          embedding_dim: 768
          return_embedding: false
          method:
          mappings:
          settings:
          create_index: true
          http_auth:
          use_ssl:
          verify_certs:
          timeout:

connections:
- sender: JinaDocumentImageEmbedder.documents
  receiver: DocumentWriter.documents

max_runs_per_component: 100

metadata: {}

inputs:
  documents:
  - JinaDocumentImageEmbedder.documents

Parameters

Inputs

Parameter	Type	Description
documents	List[Document]	A list of Documents to embed. Each document must have an image file path in its metadata under the field specified by file_path_meta_field.

Outputs

Parameter	Type	Description
documents	List[Document]	Documents with their embedding field populated and an embedding_source entry added to their metadata.

Init Parameters

These are the parameters you can configure in Pipeline Builder:

Parameter	Type	Default	Description
api_key	Secret	Secret.from_env_var('JINA_API_KEY')	The Jina API key.
model	str	jina-clip-v2	The name of the Jina multimodal model. Supported models: jina-clip-v1, jina-clip-v2, jina-embeddings-v4. See the Jina documentation for the full list.
base_url	str	https://api.jina.ai/v1/embeddings	The Jina API base URL.
file_path_meta_field	str	file_path	The document metadata field that contains the file path to the image or PDF.
root_path	Optional[str]	None	The root directory for resolving relative file paths. When None, file paths are treated as absolute.
embedding_dimension	Optional[int]	None	Number of embedding dimensions to return. Only supported by jina-embeddings-v4.
image_size	Optional[tuple[int, int]]	None	Target dimensions (width, height) to resize images while maintaining aspect ratio. Reduces memory usage and processing time.
batch_size	int	5	Number of images to send in each API request.

Run Method Parameters

These are the parameters you can configure for the component's run() method. This means you can pass these parameters at query time through the API, in Playground, or when running a job. For details, see Modify Pipeline Parameters at Query Time.

Parameter	Type	Description
documents	List[Document]	A list of Documents to embed.

Related Information

• JinaDocumentEmbedder
• Pipeline Components