I have a Deconvolution layer which bellows a Convolution layer. The convolution layer has 32 feature maps. Hence, the input of the deconvolution is 32 feature maps. The deconvolution layer is used as an upscore layer in segmentation. Do I need set the num_output in Deconvolution layer as 32? Or I can choose any number what I can?
layer {
name: "Convolution1"
type: "Convolution"
bottom: "data"
top: "Convolution1"
convolution_param {
num_output: 32
bias_term: false
kernel_size: 3
stride: 1
weight_filler {
type: "msra"
}
}
}
layer {
name: "Deconvolution1"
type: "Deconvolution"
bottom: "Convolution1"
top: "Deconvolution1"
convolution_param {
num_output: 32
pad: 1
kernel_size: 10
stride: 2
}
}