def backward(ctx, grad_output):
input1, input2 = ctx.saved_tensors
with torch.cuda.device_of(input1):
rbot1 = input1.new()
rbot2 = input2.new()
grad_input1 = input1.new()
grad_input2 = input2.new()
correlation_cuda.backward(
input1,
input2,
rbot1,
rbot2,
grad_output,
grad_input1,
grad_input2,
ctx.pad_size,
ctx.kernel_size,
ctx.max_displacement,
ctx.stride1,
ctx.stride2,
ctx.corr_multiply,
)
return grad_input1, grad_input2, None, None, None, None, None, None
def backward(self, grad_output):
input1, input2 = self.saved_tensors
with torch.cuda.device_of(input1):
rbot1 = input1.new()
rbot2 = input2.new()
grad_input1 = input1.new()
grad_input2 = input2.new()
correlation_cuda.backward(input1, input2, rbot1, rbot2, grad_output, grad_input1, grad_input2,
self.pad_size, self.kernel_size, self.max_displacement,self.stride1, self.stride2, self.corr_multiply)
return grad_input1, grad_input2
def backward(ctx, grad_output):
input1, input2 = ctx.saved_tensors
pad, kernel, max_d, stride1, stride2, corr_m = ctx.arg
with torch.cuda.device_of(input1):
rbot1 = input1.new()
rbot2 = input2.new()
grad_input1 = input1.new()
grad_input2 = input2.new()
correlation_cuda.backward(input1, input2, rbot1, rbot2, grad_output, grad_input1, grad_input2,
pad, kernel, max_d, stride1, stride2, corr_m)
return grad_input1, grad_input2
def backward(self, grad_output):
input1, input2 = self.saved_variables
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
grad_input1, grad_input2 = correlation_cuda.backward(
input1,
input2,
grad_output,
kH,
kW,
patchH,
patchW,
padH,
padW,
dilation_patchH,
dilation_patchW,
dH,
dW,
)
return grad_input1, grad_input2
def backward(ctx, grad_output):
input1, input2 = ctx.saved_tensors
pad_size, kernel_size, max_displacement, stride1, stride2, corr_multiply = 3, 3, 20, 1, 2, 1
with torch.cuda.device_of(input1):
rbot1 = input1.new()
rbot2 = input2.new()
grad_input1 = input1.new()
grad_input2 = input2.new()
#correlation_cuda.backward(input1, input2, rbot1, rbot2, grad_output, grad_input1, grad_input2,
# self.pad_size, self.kernel_size, self.max_displacement,self.stride1, self.stride2, self.corr_multiply)
correlation_cuda.backward(input1, input2, rbot1, rbot2,
grad_output, grad_input1, grad_input2,
ctx.pad_size, ctx.kernel_size,
ctx.max_displacement, ctx.stride1,
ctx.stride2, ctx.corr_multiply)
return grad_input1, grad_input2, None, None, None, None, None, None
def backward(ctx, grad_output):
input1, input2, = ctx.saved_varaibles
pad_size = ctx.pad_size
kernel_size = ctx.kernel_size
max_displacement = ctx.max_displacement
stride1 = ctx.stride1
stride2 = ctx.stride2
corr_multiply = ctx.corr_multiply
with torch.cuda.device_of(input1):
rbot1 = input1.new()
rbot2 = input2.new()
grad_input1 = input1.new()
grad_input2 = input2.new()
correlation_cuda.backward(input1, input2, rbot1, rbot2,
grad_output, grad_input1, grad_input2,
pad_size, kernel_size, max_displacement,
stride1, stride2, corr_multiply)
return grad_input1, grad_input2
