Hi @MrGiovanni , in your built_xnet function, you create the model as follow
_x = Conv2D(classes, (3,3), padding='same', name='final_conv')(interm[n_upsampleblocks])
x = Activation(activation, name=activation)(x)
model = Model(input, x)
My problem:
Is _interm[n_upsampleblocks] the combining of X(0,1), X(0,2),X(0,3),X(0,4) as specified in deep supervision? Having studied your algorithm, I noticed that _interm[n_upsampleblocks] is just up_block(X(0,4)) which is not normal.
Creating the model at this level, how do you benefit from the MODEL PRUNING as you indicate in the paper?
Thanks
Hi @MrGiovanni , in your built_xnet function, you create the model as follow _x = Conv2D(classes, (3,3), padding='same', name='final_conv')(interm[n_upsampleblocks]) x = Activation(activation, name=activation)(x) model = Model(input, x)
My problem: