umuthopeyildirim's picture
here we go
bd86ed9
raw
history blame contribute delete
658 Bytes
import torch
import torch.nn as nn
import numpy as np
class Sum_depth(nn.Module):
def __init__(self):
super(Sum_depth, self).__init__()
self.sum_conv = nn.Conv2d(1, 1, kernel_size=3, stride=1, padding=1, bias=False)
sum_k = np.array([[1, 1, 1], [1, 1, 1], [1, 1, 1]])
sum_k = torch.from_numpy(sum_k).float().view(1, 1, 3, 3)
self.sum_conv.weight = nn.Parameter(sum_k)
for param in self.parameters():
param.requires_grad = False
def forward(self, x):
out = self.sum_conv(x)
out = out.contiguous().view(-1, 1, x.size(2), x.size(3))
return out