Classes
The following classes are available globally.
-
Holds the base options used for creation of any MediaPipe task, including model asset path and hardware delegate configuration.
Declaration
Swift
class BaseOptions : NSObject, NSCopying -
Represents the embedding extracted by an embedder task head.
Declaration
Swift
class Embedding : NSObject -
Represents the embedding results of a model.
Declaration
Swift
class EmbeddingResult : NSObject -
In-memory vector store implementation backed by an ephemeral
:memory:SQLite vector database.Declaration
Swift
class InMemoryVectorStore : SqliteVectorStore -
Represents a single record stored in or retrieved from a vector store.
Declaration
Swift
class RetrievalRecord : NSObject -
Represents a single ranked result retrieved from
SemanticRetriever.Declaration
Swift
class RetrievalResult : NSObject -
Performs on-device semantic search, vector storage, and query retrieval across text, images, and audio.
Declaration
Swift
class SemanticRetriever : NSObject -
Configuration container (also bridged as
SemanticRetrieverComponents) for initializingSemanticRetriever.Declaration
Swift
class SemanticRetrieverOptions : NSObject -
Vector store implementation backed by a persistent SQLite database using the native SQLite vector engine.
Declaration
Swift
class SqliteVectorStore : NSObject -
Represents a modality part of a multi-modal content block (subclassed by
TextPart,ImagePart, andAudioPart).Declaration
Swift
class TaskPart : NSObject -
Performs multi-modal embedding extraction on text, images, audio, or composite content blocks using the native LiteRT-LM EmbeddingEngine.
Declaration
Swift
class UniversalEmbedder : NSObject -
Options for setting up a
UniversalEmbedder.Declaration
Swift
class UniversalEmbedderOptions : TaskOptions, NSCopying -
Represents the multi-modal embedding result generated by
UniversalEmbedder.Declaration
Swift
class UniversalEmbedderResult : NSObject -
Represents a vector entry stored inside a vector store.
Declaration
Swift
class VectorStoreRecord : NSObject