msc-buaa / keras-progressive_growing_of_gans Goto Github PK
View Code? Open in Web Editor NEWKeras implementation of Progressive Growing of GANs for Improved Quality, Stability, and Variation.
License: MIT License
Keras implementation of Progressive Growing of GANs for Improved Quality, Stability, and Variation.
License: MIT License
The program doesn't read the path to the celebA dataset correctly in Windows, but I've gotten it to work correctly with a few tweaks ; please look for a pull request from me soon if interested.
When I am trying to run
python h5tool.py create_celeba_channel_last face_dataset.h5 /home/uzzal/dataset/celeba/img_align_celeba
I got this error --
Traceback (most recent call last): File "h5tool.py", line 259, in <module> execute_cmdline(sys.argv) File "h5tool.py", line 241, in execute_cmdline p.add_argument( 'h5_filename', help='HDF5 file to inspect') UnboundLocalError: local variable 'p' referenced before assignment
you just statically initialize scaled weight. But scale should be done in runtime
my correct impl based on original repo:
class WScaleConv2DLayer(KL.Conv2D):
def __init__(self, *args, **kwargs):
kwargs['kernel_initializer'] = keras.initializers.random_normal()
super(WScaleConv2DLayer,self).__init__(*args,**kwargs)
def build(self, input_shape):
super().build(input_shape)
kernel_shape = K.int_shape(self.kernel)
std = np.sqrt(2) / np.sqrt( np.prod(kernel_shape[:-1]) )
self.wscale = K.constant(std, dtype=K.floatx() )
def call(self, input, **kwargs):
k = self.kernel
self.kernel = self.kernel*self.wscale
x = super().call(input,**kwargs)
self.kernel = k
return x
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.