Skip to content

Source codes for "Automatic Font Group Recognition in Early Printed Books" @ DHd 2019

Notifications You must be signed in to change notification settings

seuretm/typegroups-classification-projection

Repository files navigation

typegroups-classification-projection

About data

To run this code, you will need to have document images in the following structure:

a base directory
  Containing at least one sub-directory
    Containing at least one image

The sub-directories names correspond to the class you attributed to the document images they contain.

Programs descriptions

feature_extraction

Extracts values produces by the penultimate layer of the neural network from a given amount of patches from each page, and stores them into a file.

python3 feature_extraction.py classifier patch_count input_folder output_file

project_features

Computes a t-SNE from previously computed features, plots it, and stores it as an image.

patchwise_classification

Produces classification scores for individual patches can be obtained with patchwise_classification, and stores the whole as an html page. Note that a folder called "img" will be needed, otherwise the execution will fail.

About

Source codes for "Automatic Font Group Recognition in Early Printed Books" @ DHd 2019

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages