Theoretically yes. There are two parts to the application code, the Image analysis and the Neural Network. The NN part can handle pretty much anything you can throw at it. It just needs the learning data in a consistent input form fed to it. That is where the image analysis part comes in. My application is done to demonstrate the relevance of Neural Networks, and hence lacks a robust image analysis implementation. In fact it can only work with artificially generated bitmap files to simulate input image. If the image has any level of noise, the application fails to be useful. So if you want to process license plates or other types of input with a high signal to noise ratio, I suggest you work first with a highly robust image analysis algorithm and then try to integrate NNs into it.
It has been a long time since I worked with the code and hence I can not be of much help to you in any meaningful way.
i test this sentence(Go Swimming)in Latin Times Roman
with 72pt. font size,
but the output was wo wmoomzg,
and as we know the network which we work with it was trainng,
can you hlep me please and hurry up