Add ml-traffic-sign-training

[mpdmanash]

Machine Learning codes in python to train a three layer neural network
and two methods of TensorFlow library on German traffic sign dataset.

[krishna95]

@ManashRaja, Good Work on the models. Couple of things. The code for the tensorflow1.py does not compile. Also send a link for the dataset and try to incorporate it in the code itself. Also change the names of tensorflow1.py and tensorflow2.py to appropriate names.
@abinashmeher999 Can you please look at the PR and suggest changes.
@kumarkrishna Can you please comment on the tensorflow2.py which involves the training using Deep MNIST features of tensorflow. I consider it to be heavily overfitted.

[abinashmeher999]

Add this in requirements.txt.

[abinashmeher999]

Looks good. Nicely done. Please make time to document your functions. It will be a huge help to both you and others who will be referring to this code.

[abinashmeher999]

And this one too. I am surprised this wasn't in the requirements till now.

[hargup]

Some nitpicks,

• Conventionally we write import numpy as np instead of import numpy as Nsticking to the convention makes things easier and I also suggest that you adapt PEP8 style guide https://www.python.org/dev/peps/pep-0008/
• In create_processed_dataset.py and create_dataset.py you have used tabs instead of spaces, it will mess things up when someone who uses spaces for indentation will edit the same file. You don't need to press space, you should configure your editor to expand a tab to four space, search around the web you should find how to do it, if you don't come back asking for help.

[mpdmanash]

Thanks for all the suggestions. I will do the necessary changes and update the PR.
@abinashmeher999 I didn't know about the requirements.txt. Will learn and make the file. Also, shall I include the dataset images? Or should I just provide the download links for them from its original source?

[abinashmeher999]

@ManashRaja https://pip.pypa.io/en/stable/reference/pip_freeze/ See the 2nd example.