A deeplearning4j based captcha solver for for Taiwan Stock Exchange's website.
Compared with my another project twse-captcha-solver-java, which was based on OCR, this captcha solver will have more accurate result.
This project is built with three subprojects:
-
A simple captcha image downloader based on selenium and chromedriver. This is the first step to build your own datasets.
-
After collect enough captcha images, feed these datasets to this project and get the training model.
-
Use the training model generated from
twse-captcha-machine-learning
and create your own captcha solver.
For more info, please see each project's README.
Yes, this project is part of my side project. However, you should build your own dataset to training the captcha solver.
No, the datasets I have will only be released public after Taiwan Stock Exchange's website change their's captcha algorithm.
Currently I have no plan to release these datasets, if you really can't build your own, you can ask for commercial support
:).
Just for fun :P
Copyright © 2018 Yen-Chin, Lee <coldnew.tw@gmail.com>
Distributed under the MIT License.