How to fix error with pytorch conv2d function?-CodePudding

I am trying to use conv2d function on these two tensors:

Z = np.random.choice([0,1],size=(100,100))
Z = torch.from_numpy(Z).type(torch.FloatTensor)

print(Z)

tensor([[0., 0., 1.,  ..., 1., 0., 0.],
        [1., 0., 1.,  ..., 1., 1., 1.],
        [0., 0., 0.,  ..., 0., 1., 1.],
        ...,
        [1., 0., 1.,  ..., 1., 1., 1.],
        [1., 0., 1.,  ..., 0., 0., 0.],
        [0., 1., 1.,  ..., 1., 0., 0.]

and

filters = torch.tensor(np.array([[1,1,1],
                        [1,0,1],
                        [1,1,1]]), dtype=torch.float32)

print(filters)

tensor([[1., 1., 1.],
        [1., 0., 1.],
        [1., 1., 1.]])

But when I try to do torch.nn.functional.conv2d(Z,filters) this error returns:

RuntimeError: weight should have at least three dimensions

I really don't understand what is the problem here. How to fix it?

CodePudding user response：

The input to torch.nn.functional.conv2d(input, weight) should be

You can use unsqueeze() to add fake batch and channel dimensions thus having sizes: input: (1, 1, 100, 100) and weight: (1, 1, 3, 3).

torch.nn.functional.conv2d(Z.unsqueeze(0).unsqueeze(0), filters.unsqueeze(0).unsqueeze(0))