Method __init__

Selective Kernel Convolution Module As described in Selective Kernel Networks (https://arxiv.org/abs/1903.06586) with some modifications. Largest change is the input split, which divides the input channels across each convolution path, this can be viewed as a grouping of s

(self, in_channels, out_channels=None, kernel_size=None, stride=1, dilation=1, groups=1,
                 rd_ratio=1./16, rd_channels=None, rd_divisor=8, keep_3x3=True, split_input=True,
                 act_layer=nn.ReLU, norm_layer=nn.BatchNorm2d, aa_layer=None, drop_layer=None)