def forward(self, input1, input2):
self.save_for_backward(input1, input2)
kH, kW = self.kernel_size
patchH, patchW = self.patch_size
padH, padW = self.padding
dilation_patchH, dilation_patchW = self.dilation_patch
dH, dW = self.stride
output = correlation.forward(input1, input2,
kH, kW, patchH, patchW,
padH, padW, dilation_patchH, dilation_patchW,
dH, dW)
return output
def forward(self, input1, input2):
self.save_for_backward(input1, input2)
kH, kW = self.kernel_size
patchH, patchW = self.patch_size
padH, padW = self.padding
dilation_patchH, dilation_patchW = self.dilation_patch
dH, dW = self.stride
if input1.type() != 'torch.cuda.FloatTensor':
input1 = input1.type('torch.cuda.FloatTensor')
input2 = input2.type('torch.cuda.FloatTensor')
output = correlation.forward(input1, input2, kH, kW, patchH, patchW,
padH, padW, dilation_patchH,
dilation_patchW, dH, dW)
if output.type() != 'torch.cuda.HalfTensor':
output = output.type('torch.cuda.HalfTensor')
return output
def forward(ctx,
input1,
input2,
kernel_size=1,
patch_size=1,
stride=1,
padding=0,
dilation_patch=1):
ctx.save_for_backward(input1, input2)
kH, kW = ctx.kernel_size = _pair(kernel_size)
patchH, patchW = ctx.patch_size = _pair(patch_size)
padH, padW = ctx.padding = _pair(padding)
dilation_patchH, dilation_patchW = ctx.dilation_patch = _pair(
dilation_patch)
dH, dW = ctx.stride = _pair(stride)
output = correlation.forward(input1, input2, kH, kW, patchH, patchW,
padH, padW, dilation_patchH,
dilation_patchW, dH, dW)
return output