no such layer flatten

[chaturbhuj02]

File "", line 1, in
data_manager.preprocess()

File "C:/Users/chaturbhuj/Downloads/Compressed/neural_image_captioning-master/src/data_manager.py", line 75, in preprocess
self.get_image_features(self.image_directory)

File "C:/Users/chaturbhuj/Downloads/Compressed/neural_image_captioning-master/src/data_manager.py", line 232, in get_image_features
output=base_model.get_layer('flatten').output)

File "C:\Users\chaturbhuj\Anaconda3\lib\site-packages\keras\engine\topology.py", line 1868, in get_layer
raise ValueError('No such layer: ' + name)

ValueError: No such layer: flatten

[chaturbhuj02]

could you help me with that error

[vasudhabansal1994]

I am also facing the same error. Could you please help me to solve this issue.
Thank you.

[jubayer-hasan]

I'm having this error also.
File "train.py", line 22, in
data_manager.preprocess()
File "/home/jubayer/data_manager.py", line 75, in preprocess
self.get_image_features(self.image_directory)
File "/home/jubayer/data_manager.py", line 232, in get_image_features
output=base_model.get_layer('flatten').output)
File "/home/jubayer/Jubayer/lib/python3.5/site-packages/keras/engine/topology.py", line 1887, in get_layer
raise ValueError('No such layer: ' + name)
ValueError: No such layer: flatten
How can I solve this?

[HankyuJang]

Did anyone solve this error? I am stuck here...

[jubayer-hasan]

I think there was a problem extracting the image feature. If you set the flag to 'True' then it extracts all image features from scratch. But if you don't and run the train file then it uses the pre trained model of image features given by the author to generation. So, only the data_manager.preprocess() has the problem here.

[devinhee]

before you get the base_model, you can check the model summary to see name of each layer.
like this:
base_model = resnet50.ResNet50(weights="imagenet", include_top=True) base_model.summary()
you will find whether there is a layer called 'flatten', and find out name of the layer you real need
and then output=base_model.get_layer('flatten').output)