/TextFontClassification

Primary LanguagePython

Text Detection and font classification

Texts are detected using contours and then a svm is trained on HOG features for classifying each text into multiple(10) fonts.

dependencies(worked on):

  1. python 3.7.5
  2. numpy 1.19.5
  3. PIL 6.1.0
  4. skimage 0.18.1
  5. sklearn 0.24.1

run test:

  1. python font_detector.py <path/to/image.jpg> -test
  2. python font_detector.py <path/to/image.jpg> -show
  3. python font_detector.py <path/to/image/folder> -test
  4. python font_detector.py <path/to/image/folder> -show

generate data

  1. python gen_data.py
  2. python gen_data.py <num_samples_to_generate>

run eval:

  1. python font_detector.py -eval

run train:

  1. python font_detector.py -train

@author: Danish Ansari

@date: 2021/01/25