Keras: TypeError: __call__() got an unexpected keyword argument 'name'

Question

I am attempting to get DeepCell to work with keras. I'm running keras 2, but I added in the UID code from keras 1 to let DeepCell run. other than that I just downloaded the dependencies and software files, and I'm getting the following error.

Traceback:

  Traceback (most recent call last):
  File "/home/birtwistlelab/DeepCell/keras_version/running_template.py",line 65, in <module>
cytoplasm_predictions = run_models_on_directory(data_location,cyto_channel_names, cyto_location, model_fn = cyto_fn,list_of_weights = list_of_cyto_weights, image_size_x = image_size_x, image_size_y = image_size_y,win_x = win_cyto, win_y = win_cyto, std = False, split = False)
 File "/home/birtwistlelab/DeepCell/keras_version/cnn_functions.py", line 1491, in run_models_on_directory
model = model_fn(batch_input_shape = batch_input_shape, n_features = n_features, weights_path = list_of_weights[0])
File "/home/birtwistlelab/DeepCell/keras_version/model_zoo.py", line 528, in sparse_bn_feature_net_61x61
model.add(sparse_Convolution2D(64, 3, 3, d = d, init = init,  batch_input_shape = batch_input_shape, border_mode='valid', W_regularizer = l2(reg)))
File "/home/birtwistlelab/miniconda2/lib/python2.7/site-packages/keras/models.py", line 436, in add
layer(x)
File "/home/birtwistlelab/miniconda2/lib/python2.7/site-packages/keras/engine/topology.py", line 569, in __call__
self.build(input_shapes[0])
File "/home/birtwistlelab/DeepCell/keras_version/cnn_functions.py", line 1012, in build
self.W = self.init(self.W_shape, name='{}_W'.format(self.name))
TypeError: __call__() got an unexpected keyword argument 'name'

Code I am trying to run:

"""
Load data
"""
direc_name = "/home/birtwistlelab/DeepCell/validation_data/HeLa/"
data_location = os.path.join(direc_name, 'RawImages')
cyto_location = os.path.join(direc_name, 'Cytoplasm')
nuclear_location = os.path.join(direc_name, 'Nuclear')
mask_location = os.path.join(direc_name, 'Masks')

cyto_channel_names = ['phase','farred']
nuclear_channel_names = ['farred']

trained_network_cyto_directory = "/home/birtwistlelab/DeepCell/trained_networks/HeLa/"
trained_network_nuclear_directory = "/home/birtwistlelab/DeepCell/trained_networks/Nuclear/"

cyto_prefix = "2017-06-21_HeLa_all_61x61_bn_feature_net_61x61_"
nuclear_prefix = "2016-07-12_nuclei_all_61x61_bn_feature_net_61x61_"

win_cyto = 30
win_nuclear = 30

image_size_x, image_size_y = get_image_sizes(data_location, nuclear_channel_names)
image_size_x /= 2
image_size_y /= 2

"""
Define model
"""

list_of_cyto_weights = []
for j in xrange(5):
    cyto_weights = os.path.join(trained_network_cyto_directory,  cyto_prefix + str(j) + ".h5")
    list_of_cyto_weights += [cyto_weights]

list_of_nuclear_weights = []
for j in xrange(5):
    nuclear_weights = os.path.join(trained_network_nuclear_directory,  nuclear_prefix + str(j) + ".h5")
    list_of_nuclear_weights += [nuclear_weights]

 print list_of_nuclear_weights

 """
 Run model on directory
 """

 cytoplasm_predictions = run_models_on_directory(data_location, cyto_channel_names, cyto_location, model_fn = cyto_fn,list_of_weights = list_of_cyto_weights, image_size_x = image_size_x, image_size_y = image_size_y,win_x = win_cyto, win_y = win_cyto, std = False, split = False)

nuclear_predictions = run_models_on_directory(data_location, nuclear_channel_names, nuclear_location, model_fn = nuclear_fn,list_of_weights = list_of_nuclear_weights, image_size_x = image_size_x, image_size_y = image_size_y,win_x = win_nuclear, win_y = win_nuclear, std = False, split = False)



"""
Refine segmentation with active contours
"""

    nuclear_masks = segment_nuclei(nuclear_predictions, mask_location = mask_location, threshold = 0.75, area_threshold = 100, solidity_threshold = 0.75, eccentricity_threshold = 0.95)
cytoplasm_masks = segment_cytoplasm(cytoplasm_predictions, nuclear_masks = nuclear_masks, mask_location = mask_location, smoothing = 1, num_iters = 120)


"""
Compute validation metrics (optional)
"""
direc_val = os.path.join(direc_name, 'Validation')
imglist_val = nikon_getfiles(direc_val, 'validation_interior')
val_name = os.path.join(direc_val, imglist_val[0]) 
print val_name
val = get_image(val_name)
val = val[win_cyto:-win_cyto,win_cyto:-win_cyto]
cyto = cytoplasm_masks[0,win_cyto:-win_cyto,win_cyto:-win_cyto]
nuc = nuclear_masks[0,win_cyto:-win_cyto,win_cyto:-win_cyto]
print val.shape, cyto.shape, nuc.shape
dice_jaccard_indices(cyto, val, nuc)

I'm using the https://covertlab.github.io/DeepCell/starting/ guide. I've think it probably has something to due with the versions of the dependencies, as most of the other problems I was having were, but I'm not entirely sure. Any help would be greatly appreciated, is any other information is needed to help, let me know.

Solution

DeepCell (at least this version,) is not fully compatible with Keras 2.

• [GitHub]: CovertLab/DeepCell - (master) DeepCell/keras_version/cnn_functions.py:

  • Line #971 (part of __init__):

    d = 1, init='glorot_uniform', activation='linear', weights=None,
    

  • Line #982:

    self.init = initializations.get(init)
    

  • Line #1012:

    self.W = self.init(self.W_shape, name='{}_W'.format(self.name))
    

• [GitHub]: keras-team/keras - (keras-1) keras/keras/initializations.py:56:

  def glorot_uniform(shape, name=None, dim_ordering='th'):
  

• [GitHub]: keras-team/keras - (keras-2) keras/keras/initializers.py:325:

  def glorot_uniform(seed=None):
  

So, in Keras 2 the initializers headers were (drastically) changed, and DeepCell code doesn't reflect that. In order to fix your error, either:

• Switch to Keras 1
• Modify DeepCell code to be Keras 2 compatible. But I seriously doubt that cnn_functions.py:1012 would be the only place to modify