As far as I have searched, each individual digits can be detected by using MNIST Dataset and CNN. But for Detecting multiple digits in a image including points as in decimal points, i don't know which approach I have to take.
My real case situation is to develop a model that can convert a image containing a digits in a digital scenario - like an image of an digital energy meter, or calculator, or digital gauges used in manufacturing industries.
I just want to know, for this use case is creating a model for recognition is a correct approach. Or another simple method is available.
End Result is that I want to run an application in smartphone so i can do this natively. I have tried OCR, but its accuracy is varying and in smartphone OpenCV is heavy to render.
Any help would be appreciated,