DenseNet

DenseNet copied to clipboard

I am trying to convert the densenet161 model to frozen graph(pb) or checkpoint, but after loading the model and while saving the checkpoint file I am getting the below error....

venkat-kittu

#from keras.backend import normalize_data_format from keras.utils.conv_utils import normalize_data_format should replace first line of code to the second in subpixel.py to run.

wubinbai

Poor CIFAR100 accuracy

1

Hello, I tried the cifar100.py with the default setting, only got around 40% accuracy. Then I tried (depth=100, epochs=300, bottleneck=True), only got 55% accuracy. Could you please let me know...

christa60

I saw this error after making the model on imagenet What should I do? 

alavi8

About Densenet architecture

1

How is L(L+1)/2 implemented?

thaingiW

Plans for memory efficient implementation in Keras?

1

Hi @titu1994, I checked out @joeyearsleys 's [repo](https://github.com/joeyearsley/efficient_densenet_tensorflow) but its limited to densenet-121 architecture, while your's has other architectures supported as well. Are you planning to add memory efficiency support...

aayushARM

Update densenet.py

9

exclude the input with the feature maps in the concatenation after the last desnblock

jrwin