Keras Implementation of The One Hundred Layers Tiramisu: Fully Convolutional DenseNets for Semantic Segmentation by (Simon Jégou, Michal Drozdzal, David Vazquez, Adriana Romero, Yoshua Bengio)
Nice github repository.
I am trying your 103-layer tiramisu code. It seems to me that there is no skip-connections, or maybe I have misunderstood something.
Furthermore. When I train the model in keras the model summary displays 428,328,755 model parameters.
As far as I am informed the original tiramisu 103 layer uses 9.4 million paramers. Maybe I am misunderstanding something. I hope you can clarify it for me.
Hello
Nice github repository. I am trying your 103-layer tiramisu code. It seems to me that there is no skip-connections, or maybe I have misunderstood something. Furthermore. When I train the model in keras the model summary displays 428,328,755 model parameters. As far as I am informed the original tiramisu 103 layer uses 9.4 million paramers. Maybe I am misunderstanding something. I hope you can clarify it for me.
best regards