HemingwayLee / deepfont-implement

deepfont-implement

• How to identify fonts in pictures using deep learning? we can do visual font recognition (VFR) using DeepFont algorithm (from Adobe). The paper is here

• This implementation is from Font_Recognition-DeepFont, in this repo
  • requirements.txt and Dockerfile are added so we can set it up easily
  • Github action is added so we can make sure Dockerfile do work
  • Data generation script create_data.py is provided
  • train.py and eval.py are provided so we can run training and evaluation without jupyter notebook

How to run it locally

• setup

python3 -m "virtualenv" venv
source venv/bin/activate
pip3 install -r requirements.txt

• create dataset
  • -c: the number of generated images for each font in fonts folder
  • -f: the target folder

python3 create_data.py -c 1 -f data

• train
  • -e: epoch

python3 train.py -e 1

• evaluate
  • -d: the path of testing file
  • -m: model filename

python3 eval.py -d test_data/Times.jpg -m top_model.h5

Run by docker

• Build Dockerfile

docker build -t mydeepfont .

• Run dockerfile
  • access http://127.0.0.1:8888 by browser

docker run -it --rm -p8888:8888 mydeepfont