AttributeError: 'str' object has no attribute 'decode'

[LZZ383]

#from keras_vggface.vggface import VGGFace

VGGFace ResNet50

#vggface = VGGFace(include_top=False, model='resnet50', input_shape=(224, 224, 3))'

from colab_demo.vggface_models import RESNET50
vggface = RESNET50(include_top=False, weights=None, input_shape=(224, 224, 3))
vggface.load_weights("rcmalli_vggface_tf_notop_resnet50.h5")

#from keras.applications.resnet50 import ResNet50
#vggface = ResNet50(include_top=False, input_shape=(224, 224, 3))

#vggface.summary()

model.build_pl_model(vggface_model=vggface, before_activ=loss_config["PL_before_activ"])
model.build_train_functions(loss_weights=loss_weights, **loss_config)

Error:

AttributeError Traceback (most recent call last)
in ()
6 from colab_demo.vggface_models import RESNET50
7 vggface = RESNET50(include_top=False, weights=None, input_shape=(224, 224, 3))
----> 8 vggface.load_weights("rcmalli_vggface_tf_notop_resnet50.h5")
9
10 #from keras.applications.resnet50 import ResNet50

1 frames
/usr/local/lib/python3.7/dist-packages/keras/engine/topology.py in load_weights_from_hdf5_group(f, layers, reshape)
3326 """
3327 if 'keras_version' in f.attrs:
-> 3328 original_keras_version = f.attrs['keras_version'].decode('utf8')
3329 else:
3330 original_keras_version = '1'

AttributeError: 'str' object has no attribute 'decode'

[skiwheelr]

This seems to be an issue with the upgrade to tensorflow 2.0

https://stackoverflow.com/questions/58574610/python3-recognizes-tensorflow-but-doesnt-recognize-any-of-its-attributes

I can see what the problem is, but i'm not sure how to rectify it.

[skiwheelr]

There is an open PR that should fix some of these.

[mushfiq-ncsu]

I solved this in this way: (tested multiple time on different machines)

pip install 'h5py==2.10.0' --force-reinstall