FPN Lateral Parameter Count

Problem

Count the parameters of the 1x1 lateral convolutions in an FPN.

Each level has a c_in to fpn_dim convolution with a bias, contributing c_in * fpn_dim + fpn_dim parameters. Return the total over all backbone channels.

Test cases

case 1

in: [[256,512,1024,2048],256]

expected: 984064

case 2

in: [[64],16]

expected: 1040

case 3

in: [[32,64,128],32]

expected: 7264

Starter codePython

def fpn_lateral_params(backbone_channels, fpn_dim):
    # Your code here
    pass

3 test cases · run your code against them in the editor