Task

Cross-Modal Retrieval

Canonical ID https://id.searchplex.net/cross-modal-retrieval/

A retrieval task in which the query and retrieved items may be expressed in different modalities, such as retrieving images from text queries or text from image queries.

Also called

  • Cross-Modality Retrieval
  • Cross-Modal Search

Scope

Cross-modal retrieval is a narrower case within multimodal retrieval. It emphasizes matching across a modality boundary rather than searching within one modality or combining multiple modalities in a single representation.

Observed terms

  • Image-Text Retrieval academic
  • Text-to-Image Search industry
  • Image-to-Text Search industry
  • Cross-Type Search vendor

Part of family

Not equivalent to

  • Cross-Modal Retrieval is not Image Retrieval. Image retrieval retrieves images; cross-modal retrieval specifically matches across modalities, such as text-to-image or image-to-text.
  • Cross-Modal Retrieval is not Reverse Image Search. Reverse image search uses an image as the query, often to find visually similar or related images; cross-modal retrieval crosses between different modalities.
  • Cross-Modal Retrieval is not Vector Search. Cross-modal retrieval may use vector search, but it is a task involving modality boundaries rather than a vector-search mechanism itself.
  • Cross-Modal Retrieval is not Multimodal Retrieval. Multimodal retrieval is the broader family involving multiple modalities; cross-modal retrieval specifically crosses a modality boundary between query and retrieved items.

Learn more

ID