Closed Annie98 closed 4 years ago
Tensorflow 1.12+ should be fine. Not tested with 2.0, but latest Keras from pypi does support Tensorflow 2 so I expect it to work.
Many thanks for your replying. I use tensorflow 1.15.0 and keras 2.3.1, but with the problem as follow:
/home/anni/anaconda3/lib/python3.6/site-packages/h5py/init.py:36: FutureWarning: Conversion of the second argument of issubdtype from float
to np.floating
is deprecated. In future, it will be treated as np.float64 == np.dtype(float).type
.
from ._conv import register_converters as _register_converters
Using TensorFlow backend.
Traceback (most recent call last):
File "/home/anni/code/Contouring/Code/Models/coordUnet.py", line 119, in
def cdunet(nClasses, input_height, input_width): img_input = Input(shape=(input_height, input_width, 1)) x = CoordinateChannel2D()(img_input) ft = Conv2D(filters=3, kernel_size=3, padding='same', kernel_initializer='glorot_normal', bias_initializer='zeros' )(x) md = Model(inputs=img_input, outputs=ft) vgg_streamlined = vgg16.VGG16(include_top=False, weights=None, input_tensor=md.output) assert isinstance(vgg_streamlined, Model) ...... ` I also try the version of tensorflow such as 1.13.2 ,1.14.0 and 1.12.0, but all of those have some problem. Could you help me? Thank you !
I don't know what the issue is here. Seems internally Keras error somewhere. I've stopped working with base Keras, and have just been moving stuff to TF.keras as an alternative.
If you'd like I could probably port this module to tf.keras instead.
Thanks, and Good Luck!
can you just list the version of the requriement of python packages, especially tensorflow?