Closed yoookoo closed 7 years ago
tdeboissiere
commented
7 years ago Which version of Keras are you using ? Which para,eters did you select to run it with ?
If using keras 2.0, it's likely there will be bugs as I have not yet ported the code
yoookoo
commented
7 years ago Thanks for your reply. The version of my Keras is 1.2.2 using theano backend. Errors happened when trying to joint training D with G.
img_dim = (3,256,256)
nb_patch = 16
img_dim_disc = 64
discriminator_model.summary()
Layer (type) Output Shape Param # Connected to
disc_input_0 (InputLayer) (None, 3, 64, 64) 0
disc_input_1 (InputLayer) (None, 3, 64, 64) 0
disc_input_2 (InputLayer) (None, 3, 64, 64) 0
disc_input_3 (InputLayer) (None, 3, 64, 64) 0
disc_input_4 (InputLayer) (None, 3, 64, 64) 0
disc_input_5 (InputLayer) (None, 3, 64, 64) 0
disc_input_6 (InputLayer) (None, 3, 64, 64) 0
disc_input_7 (InputLayer) (None, 3, 64, 64) 0
disc_input_8 (InputLayer) (None, 3, 64, 64) 0
disc_input_9 (InputLayer) (None, 3, 64, 64) 0
disc_input_10 (InputLayer) (None, 3, 64, 64) 0
disc_input_11 (InputLayer) (None, 3, 64, 64) 0
disc_input_12 (InputLayer) (None, 3, 64, 64) 0
disc_input_13 (InputLayer) (None, 3, 64, 64) 0
disc_input_14 (InputLayer) (None, 3, 64, 64) 0
disc_input_15 (InputLayer) (None, 3, 64, 64) 0
PatchGAN (Model) [(None, 2), (None, 51 6279554 disc_input_0[0][0]
disc_input_1[0][0]
disc_input_2[0][0]
disc_input_3[0][0]
disc_input_4[0][0]
disc_input_5[0][0]
disc_input_6[0][0]
disc_input_7[0][0]
disc_input_8[0][0]
disc_input_9[0][0]
disc_input_10[0][0]
disc_input_11[0][0]
disc_input_12[0][0]
disc_input_13[0][0]
disc_input_14[0][0]
disc_input_15[0][0]
merge_feat (Merge) (None, 32) 0 PatchGAN[17][0]
PatchGAN[18][0]
PatchGAN[19][0]
PatchGAN[20][0]
PatchGAN[21][0]
PatchGAN[22][0]
PatchGAN[23][0]
PatchGAN[24][0]
PatchGAN[25][0]
PatchGAN[26][0]
PatchGAN[27][0]
PatchGAN[28][0]
PatchGAN[29][0]
PatchGAN[30][0]
PatchGAN[31][0]
PatchGAN[32][0]
disc_output (Dense) (None, 2) 66 merge_feat[0][0]
Total params: 6,279,620
Trainable params: 6,275,652
Non-trainable params: 3,968Which generator do you use ? deconv or upsampling ?
yoookoo
commented
7 years ago Upsampling is used for generator.
For
python main.py 64 64
Same error happens:
Traceback (most recent call last):
File "main.py", line 71, in <module>
launch_training(**d_params)
File "main.py", line 8, in launch_training
train.train(**kwargs)
File "/home/yuankun/DeepLearningImplementations/pix2pix/src/model/train.py", line 84, in train
image_dim_ordering)
File "/home/yuankun/DeepLearningImplementations/pix2pix/src/model/models.py", line 298, in DCGAN
DCGAN_output = discriminator_model(list_gen_patch)
File "/home/yuankun/anaconda2/lib/python2.7/site-packages/keras/engine/topology.py", line 572, in __call__
self.add_inbound_node(inbound_layers, node_indices, tensor_indices)
File "/home/yuankun/anaconda2/lib/python2.7/site-packages/keras/engine/topology.py", line 635, in add_inbound_node
Node.create_node(self, inbound_layers, node_indices, tensor_indices)
File "/home/yuankun/anaconda2/lib/python2.7/site-packages/keras/engine/topology.py", line 172, in create_node
output_tensors = to_list(outbound_layer.call(input_tensors, mask=input_masks))
File "/home/yuankun/anaconda2/lib/python2.7/site-packages/keras/engine/topology.py", line 2247, in call
output_tensors, output_masks, output_shapes = self.run_internal_graph(inputs, masks)
File "/home/yuankun/anaconda2/lib/python2.7/site-packages/keras/engine/topology.py", line 2390, in run_internal_graph
computed_mask))
File "/home/yuankun/anaconda2/lib/python2.7/site-packages/keras/engine/topology.py", line 2247, in call
output_tensors, output_masks, output_shapes = self.run_internal_graph(inputs, masks)
File "/home/yuankun/anaconda2/lib/python2.7/site-packages/keras/engine/topology.py", line 2420, in run_internal_graph
shapes = to_list(layer.get_output_shape_for(computed_tensors[0]._keras_shape))
File "/home/yuankun/anaconda2/lib/python2.7/site-packages/keras/layers/core.py", line 821, in get_output_shape_for
assert input_shape[-1] and input_shape[-1] == self.input_dim
AssertionError
I'll look into it when I get the time. You can try to debug it by varying the patch size and inspecting inputs and outputs in the meantime
yoookoo
commented
7 years ago I have debugged by defining inputs and outputs but still not work. Which version of Keras are you using ? Version may be a problem. And which backend have you choosed? Theano or Tensorflow? It would be grateful if you can help solve this issue.
yoookoo
commented
7 years ago Everything is OK if I turn the backend into 'tensorflow'. So I still don't know what's wrong with 'theano'.
Something is wrong when running ./pix2pix/src/model/ function DCGAN
DCGAN_output = discriminator_model(list_gen_patch)