-
Notifications
You must be signed in to change notification settings - Fork 199
API
Basic Key tools of MONAI Label:
MONAI Label App
-
Active Learning: Currently, MONAI Label offers all the necessary callbacks needed to implement active learning techniques. However, for version 0.1, MONAI Label Apps used the simplest random sampling technique to fetch the next image. More info can be found in the Active Learning Section
-
Inference: This interface (class) allows the developer to use or define the inference task in the MONAI Label App. This means, all the inference transforms, types of inferrers (i.e. simple inferrer, sliding window inferrer, etc) and post transforms.
-
Training: This interface (class) allows the developer to use or redefine the training task for their own MONAI Label App. In this class, researchers can use their own transforms for pre-processing, post-processing, declare the validation dataset and the type of inferrer for validation time.
-
Custom transforms: In this file developers can create their custom transforms to apply on the dataset before, after, or during training the AI algorithm.
-
Main file: In this file developers can instantiate the App class/interface to create their own MONAI Label App
-
Custom requirements file: Custom requirements for the MONAI Label App. In this file, researchers have the option of specifying the external libraries use in their App (TensorFlow, Catalyst, Kornia etc)