Inference radtorch.inference

Inference

An Inference class creates a predictor object that utilizes a trained model to perform predictions over target image(s).

Parameters:

Name	Type	Description	Default
classifier	ImageClassifier	trained ImageClassifier.	required
use_best_model	bool	True to use the model with the lowest validation loss.	True
transform	list	Albumentations transformations. See Image Augmentation. See below.	False
device	str	Device to be used for training. Default: 'auto' which automtically detects GPU presence and uses it for feature extraction. Options: 'auto', 'cuda', 'cpu'.	'auto'

Using transform

By default, the Inference class utilizes the transforms specified in the train subset used to train the ImageClassifier. When this is not available, it will try to utilize transforms of the valid subset. You can specify specific transforms as needed instead.

Methods

predict(self, img_path, top_predictions='all', human=True, display_image=False, cmap='gray')

Performs predictions using Inference class

Parameters:

Name	Type	Description	Default
img_path	str	path to target image.	required
top_predictions	int or str	number of top predictions to return. Default = 'all' which returns all predictions.	'all'
human	bool	True to display predictions in human readable format.	True
display_image	bool	True to display the target image.	False
cmap	string	Name of Matplotlib color map to be used. See Matplotlib colormaps	'gray'

Returns:

Type	Description
	(list) list of predictions if human is set to False.