Python verification code identification module

You only need a few simple steps to have two universal identification modules, so that you can work unimpeded.

test picture test1.png

test picture test2.jpg

just choose one of the above parameters, and the default model_type is ModelType.OCR If the conf_path parameter is specified, the user-defined model will be used first.

Note: Because the module is too new, the third-party sources such as Ali/Tsinghua may not have updated their images, so the overseas sources are designated manually. In order to improve the installation speed of the dependencies, the dependencies can be installed by themselves in advance: TensorFlow/Numpy/OpenCV-Python/Pillow/PyYaml

Output results:

The speed of OCR and captcha recognition is basically around 15-2ms, and it may take 1 ms for a low CPU. This module only supports single-line recognition. If you need multi-line recognition, please use target detection to pre-cut the picture yourself.