mmpretrain.apis.inference_model

mmpretrain.apis.inference_model(model, *args, **kwargs)

Inference an image with the inferencer.

Automatically select inferencer to inference according to the type of model. It’s a shortcut for a quick start, and for advanced usage, please use the correspondding inferencer class.

Here is the mapping from task to inferencer:

• Image Classification: ImageClassificationInferencer

• Image Retrieval: ImageRetrievalInferencer

• Image Caption: ImageCaptionInferencer

• Visual Question Answering: VisualQuestionAnsweringInferencer

• Visual Grounding: VisualGroundingInferencer

• Text-To-Image Retrieval: TextToImageRetrievalInferencer

• Image-To-Text Retrieval: ImageToTextRetrievalInferencer

• NLVR: NLVRInferencer

Parameters:

• model (BaseModel | str | Config) – The loaded model, the model name or the config of the model.

• *args – Positional arguments to call the inferencer.

• **kwargs – Other keyword arguments to initialize and call the correspondding inferencer.

Returns:

The inference results.

Return type:

result (dict)