The base class for performing inference
Performer
(model
, transform_pipelines
:dict
=None
)
Performer.describe_pipeline_of_type
(pipeline_type
:str
)
Describes a configured transform pipeline.
Can be a string of "type", "item", or "batch".
Transforms are then described as:
Transform Name: `Resize`
Supported input types:
- `PIL.Image.Image`
- `fastai.vision.core.TensorBBox`
- `fastai.vision.core.TensorPoint`
Has different behavior on test set:
- Yes
if it is a fastai transform, else:
Transform Name: `function_name`