|---| | [ View source on GitHub](https://github.com/google-ai-edge/mediapipe/blob/master/mediapipe/tasks/python/retrieval/__init__.py) | MediaPipe Tasks Retrieval API. ## Classes [`class ActivationDataType`](https://developers.google.com/edge/api/mediapipe/python/mp/tasks/retrieval/ActivationDataType.md.txt): Activation data type for model execution. [`class AudioPart`](https://developers.google.com/edge/api/mediapipe/python/mp/tasks/retrieval/AudioPart.md.txt): An audio part in a multimodal content block. [`class ChunkingMode`](https://developers.google.com/edge/api/mediapipe/python/mp/tasks/retrieval/ChunkingMode.md.txt): Mode for text chunking matching C/C++ definitions. [`class ImagePart`](https://developers.google.com/edge/api/mediapipe/python/mp/tasks/retrieval/ImagePart.md.txt): An image part in a multimodal content block. [`class RetrievalRecord`](https://developers.google.com/edge/api/mediapipe/python/mp/tasks/retrieval/RetrievalRecord.md.txt): Retrieval result representing a single matched record. [`class RetrievalResult`](https://developers.google.com/edge/api/mediapipe/python/mp/tasks/retrieval/RetrievalResult.md.txt): Retrieval result representing a single matched record. [`class SemanticRetriever`](https://developers.google.com/edge/api/mediapipe/python/mp/tasks/retrieval/SemanticRetriever.md.txt): Class that manages a vector index and retrieves relevant media records. [`class SemanticRetrieverOptions`](https://developers.google.com/edge/api/mediapipe/python/mp/tasks/retrieval/SemanticRetrieverOptions.md.txt): Options for the semantic retriever task. [`class TaskPartKind`](https://developers.google.com/edge/api/mediapipe/python/mp/tasks/retrieval/TaskPartKind.md.txt): Part kind enum matching C/C++ definitions. [`class TextPart`](https://developers.google.com/edge/api/mediapipe/python/mp/tasks/retrieval/TextPart.md.txt): A text part in a multimodal content block. [`class UniversalEmbedder`](https://developers.google.com/edge/api/mediapipe/python/mp/tasks/retrieval/UniversalEmbedder.md.txt): Class that performs embedding extraction on multi-modal inputs. [`class UniversalEmbedderOptions`](https://developers.google.com/edge/api/mediapipe/python/mp/tasks/retrieval/UniversalEmbedderOptions.md.txt): Options for the universal embedder task. [`class UniversalEmbedderResult`](https://developers.google.com/edge/api/mediapipe/python/mp/tasks/retrieval/UniversalEmbedderResult.md.txt): Embedding results for a given embedder model. ## Type Aliases [`ContentPart`](https://developers.google.com/edge/api/mediapipe/python/mp/tasks/retrieval/ContentPart.md.txt)