Text detection in images

问题:

I am using below sample code for text detection in images (not handwritten) using coreml and vision.

https://github.com/DrNeuroSurg/OCRwithVisionAndCoreML-Part2

In this they have used machine learning model which supports only uppercase and numbers. Where as in my project I want upper case, lower case , numbers and few of special characters (like : ,- ).

I do not have any experience in python to do required changes and generate the required .mlmodel file using train data (which again I don't have for my requirement)

Below is the link where it is given how to create .mlmodel http://www.neurosurg.de/2017/10/18/part-3-train-your-own-model-using-keras/

does anyone has such .mlmodel? or can somebody help me how to achieve this?

this is the image in which I want to detect text