Training: 1164 images, 38305 samples
Validation: 520 images, 8198 samples
- Python 3 (I used 3.7.9)
- I recommend to use venv but you can also use global pip modules
- Install all requirements in requirements.txt with
pip install -r requirements.txt
- Note that if you don't want to use GPU you don't have to install
tensorflow-gpu
- Keras GPU installation tutorial: https://deeplizard.com/learn/video/IubEtS2JAiY
- Note that if you don't want to use GPU you don't have to install
-
Download training, validation and test files:
- train file #1:
SynthText.h5
: https://drive.google.com/file/d/1h5yzTRLFu_1cVkYPOs98RuEPB3Rdc5T-/view - train file #2:
train.h5
: https://drive.google.com/file/d/10pRYujKeJ-vtkFY7aaPZVKKAeVR-OuNr/view - validation file:
SynthText_val.h5
: https://drive.google.com/file/d/1K2EABjhUEpy79wQf4Nsepmb7Ag3DQN3g/view - test file:
test.h5
: https://drive.google.com/file/d/1s5QqcBGDk8Bm1WrA00JCQHqaFyCqie4H/view
- train file #1:
-
Put the files in correct folders (look at image above)
- Put the two training files into folder called
train
inside this project- i.e.
train/SynthText.h5
- i.e.
train/train.h5
- i.e.
- Put the test file into folder called
test
inside this project- i.e.
test/test.h5
- i.e.
- Put the validation file into folder called
validation
inside the project- i.e.
validation/SynthText_val.h5
- i.e.
- Put the two training files into folder called
Link: https://drive.google.com/file/d/1k0RFXDuC4NjLrmwEZTm-DbhCxnHJV8DB/view?usp=sharing
Put in the project root folder with name: saved_model.h5
Run $ python final.py
Run $ python final-use_model.py
https://docs.google.com/document/d/1j0Q2lFW6pZKRNUjVaTEd-THx0V0PBokFPpUba-17pWA/edit?usp=sharing