Running SPIT

This file offers a few notes on running SPIT within a Python code. You might find this iPython Notebook to be the most helpful.

Load the Classifier

SPIT needs a CNN arachitecture to run. Here is the call to load the Kast classifier:

from spit.classifier import Classifier
kast = Classifier.load_kast()

Pre-Process

Assuming you have an image in memory (as a Numpy ndarray), presume to pre-process:

pdict = spit_p.original_preproc_dict()  # dict to guide the steps
images_array = spit_p.flattened_array(data, pdict) # Generates 4 flattened images

Classify

Time to classify. Here goes:

prediction, results = spit_c.get_prediction(images_array, kast)

And this will turn the output into an image_type name:

pred_type = kast.classify_dict[prediction]

And that is that.