why there is the additional convolution in the unet_model function .
conv9 = Conv2D(filters=32,
kernel_size=(3,3),
activation=‘relu’,
padding=‘same’,
# set ‘kernel_initializer’ same as above exercises
kernel_initializer=‘he_normal’)(ublock9)
Although in u-net model this convolution step is not present
Which assignment is this?
Since it’s unet, that’s probably C4 W3 A2.
convolutional network(course-4), week 3 assignment 2
Hmm its been a long time I have done that one, frankly I dont know but my intution tells me; convolutions exrtact more features and make the process run smother in our context, i.e. the segmentations learned will be better. Thats what I think without revising this lab, @TMosh what do you think?
