A set of modules to easily build a classifier in tensorflow
The machine learning pipeline has been divided into five modules
-
loader.py - loads the images files from a directory and creates a train and valid dataset.
Load test function which load a directory of test files -
augopt.py - returns a dataset with applied augmentation and various optimisation
-
arch.py - returns the model architecture
-
fitcompile.py -
- compiles the model for loss, metrics and optimiser
- fit the model
-
evaluate.py - plots the confusion metrics and saves the missclassified images
-
usage.py - shows how to use the modules and set the hyperparameters
Documentation URL : https://classifier-modules.readthedocs.io/en/latest/