766bd71d6cca7313988b02784be6d56834e8c744,kornia/filters/filter.py,,filter3D,#Any#Any#Any#Any#,102

Before Change

    input_pad: torch.Tensor = F.pad(input, padding_shape, mode=border_type)

    // convolve the tensor with the kernel.
    return F.conv3d(input_pad, tmp_kernel.expand(c, -1, -1, -1, -1), groups=c, padding=0, stride=1)

After Change

        tmp_kernel = normalize_kernel2d(tmp_kernel.view(
            bk, dk, hk * wk)).view_as(tmp_kernel)

    tmp_kernel = tmp_kernel.expand(-1, c, -1, -1, -1)

    // pad the input tensor
    depth, height, width = tmp_kernel.shape[-3:]
    padding_shape: List[int] = compute_padding([depth, height, width])
    input_pad: torch.Tensor = F.pad(input, padding_shape, mode=border_type)

    // kernel and input tensor reshape to align element-wise or batch-wise params
    tmp_kernel = tmp_kernel.reshape(-1, 1, depth, height, width)
    input_pad = input_pad.view(-1, tmp_kernel.size(0), input_pad.size(-3), input_pad.size(-2), input_pad.size(-1))

    // convolve the tensor with the kernel.
    output = F.conv3d(input_pad, tmp_kernel, groups=tmp_kernel.size(0), padding=0, stride=1)
    return output.view(b, c, d, h, w)

Instances