2 questions please: get_encoder(I) output looks weird and increase resolution ?

[thusinh1969]

def get_encoder(I):

K1  = concatenate([I,I])
new = MRU(I  ,I ,32)
I2  = AveragePooling2D(pool_size=2)(I)

K2  = concatenate([new,I2])
new = MRU(new,I2,64)
I3  = AveragePooling2D(pool_size=2)(I2)

K3  = concatenate([new,I3])
new = MRU(new,I3,128)
I4  = AveragePooling2D(pool_size=2)(I3)

K4  = concatenate([new,I4])
out = MRU(new,I4,256)
I5  = AveragePooling2D(pool_size=2)(I4)

**K5  = concatenate([out,I5])**

**encoder = Model(I, out, name='encoder');**

return encoder,K4,K3,K2,K1

Were you intended to out = K5 instead ? Otherwise what K5 is used for ?

Also if I would like to increase resolution to say 256x256x3 or higher, which layers/changes should I look at.

I also modified a bit and put in Instance normalization, let see... !

Thanks man.
Steve