- Install all requirements(if not yet) using command
pip install -r requirements.txt - download images from here (or use your own dataset), create folder images and place all dataset into 2 folders: train and test. Image name format:
answer123-randomseed.jpeg - Configure code/utils/config.py file - img_width, img_height, format, ect.
- Run code/statistics.py for analyzing your dataset (it will print all characters, whose your captchas contains)
- code/main.py - TRAIN your model
- code/test_mode.py - Test your tensorflow model
- code/to_onnx.py - Convert your model to ONNX format
- code/onnx_tester.py - Test your ONNX model
- Write your own captcha-solver-adapter. For example you can look at this file
code/tools.py - конфиг файл (файл модели, ) + тулзы
code/test_mode.py - тест модели на реальных данных code/to_onnx.py - конверт модели tesorflow в формат onnx (для использования через onnxruntime)
vk_captcha_lib - код библиотеки vk_captcha по решению капчи