Home

DCGAN renders black noise

I followed a simple DCGAN tutorial (in French) and I tried to make it work on my own dataset.

But It keeps getting in a "darkish" noise even so my images’ dataset contain lots of white. The learning did not managed to make it disapeared.

here is the result after +4.5k epochs GAN BLACK GENERATION

Link to my dataset (mp4)

Link to the GAN epochs (mp4) (Around 30sec the noise begin to take over)

The generator function seems pretty normal:

model=tf.keras.Sequential()  model.add(layers.Dense(4*4*1024, use_bias=False, input_shape=(bruit_dim,))) model.add(layers.BatchNormalization()) model.add(layers.LeakyReLU())     model.add(layers.Reshape((4, 4, 1024)))  model.add(layers.Conv2DTranspose(512, (5, 5), strides=(2, 2), padding='same', use_bias=False)) model.add(layers.BatchNormalization()) model.add(layers.LeakyReLU())  model.add(layers.Conv2DTranspose(256, (5, 5), strides=(2, 2), padding='same', use_bias=False)) model.add(layers.BatchNormalization()) model.add(layers.LeakyReLU())  model.add(layers.Conv2DTranspose(128, (5, 5), strides=(2, 2), padding='same', use_bias=False)) model.add(layers.BatchNormalization())  model.add(layers.Conv2DTranspose(64, (5, 5), strides=(2, 2), padding='same', use_bias=False)) model.add(layers.BatchNormalization())  model.add(layers.Conv2DTranspose(3, (5, 5), strides=(2, 2), padding='same', use_bias=False, activation='sigmoid'))

Thanks in advance 🙂

Asked on July 16, 2020 in Python.
Add Comment
Cancel
0 Answer(s)

Your Answer

By posting your answer, you agree to the privacy policy and terms of service.