Fix some layers for keras 2.0+

[ntuaha]

Thanks for your project. It is very helpful.
Before running your resnet152 model, I fixed some codes for keras 2.0.
According to Keras official document ( ref to: https://github.com/keras-team/keras/wiki/Keras-2.0-release-notes), The Inferface of Conv2D are changed as follows:

1. float kernel dimension arguments become a single tuple argument, kernel size. E.g. a legacy call Conv2D(10, 3, 3) becomes Conv2D(10, (3, 3))
2. subsample -> strides. Can also be set to an integer.
3. bias -> use_bias
   (... more...)

I hope it is useful to you.

Best wishes