Discover how the original dataset was an train a new one

[Alexandre Muller]

Hi Caffe Users, 

I 'm developing an android app that captures Credit Cards number using PayCard Android library. But unfortunately the cards issues by the company I work for doesn't read on it. Looking carefully on sourcecode of the SDK I could find the models of the card capture using Caffe:

PayCards_Android/sdk/src/main/assets/cardrecognizer/model/NumberLocalization at master · faceterteam/PayCards_Android (github.com)

As the authors don't replies me.. I want to know.. Is it possible to discover how the original dataset was looking only to these files? .caffemodel and .prototxt ?

In case it is.. how can I train a new model? is it a guide to follow?

Thanks 

Alexandre