import math
from typing import Tuple
import torch
import torch.nn as nn
import torch.nn.functional as F
from kornia.filters import get_gaussian_kernel2d, spatial_gradient
from kornia.geometry.conversions import pi
def get_sift_pooling_kernel(ksize: int = 25) -> torch.Tensor:
r"""Return a weighted pooling kernel for SIFT descriptor.
Args:
ksize: kernel_size.
Returns:
the pooling kernel with shape :math:`(ksize, ksize)`.
"""
ks_2: float = float(ksize) / 2.0
xc2: torch.Tensor = ks_2 - (torch.arange(ksize).float() + 0.5 - ks_2).abs() # type: ignore
kernel: torch.Tensor = torch.ger(xc2, xc2) / (ks_2 ** 2)
return kernel
def get_sift_bin_ksize_stride_pad(patch_size: int, num_spatial_bins: int) -> Tuple:
r"""Return a tuple with SIFT parameters.
Args:
patch_size: the given patch size.
num_spatial_bins: the ggiven number of spatial bins.
Returns:
ksize, stride, pad.
"""
ksize: int = 2 * int(patch_size / (num_spatial_bins + 1))
stride: int = patch_size // num_spatial_bins
pad: int = ksize // 4
out_size: int = (patch_size + 2 * pad - (ksize - 1) - 1) // stride + 1
if out_size != num_spatial_bins:
raise ValueError(
f"Patch size {patch_size} is incompatible with \
requested number of spatial bins {num_spatial_bins} \
for SIFT descriptor. Usually it happens when patch size is too small\
for num_spatial_bins specified"
)
return ksize, stride, pad
[docs]class SIFTDescriptor(nn.Module):
r"""Module which computes SIFT descriptors of given patches.
Args:
patch_size: Input patch size in pixels.
num_ang_bins: Number of angular bins.
num_spatial_bins: Number of spatial bins.
clipval:
rootsift: if ``True``, RootSIFT (Arandjelović et. al, 2012) is computed.
Returns:
SIFT descriptor of the patches with shape.
Shape:
- Input: :math:`(B, 1, \text{num_spatial_bins}, \text{num_spatial_bins})`
- Output: :math:`(B, \text{num_ang_bins * num_spatial_bins ** 2})`
Example:
>>> input = torch.rand(23, 1, 32, 32)
>>> SIFT = SIFTDescriptor(32, 8, 4)
>>> descs = SIFT(input) # 23x128
"""
def __repr__(self) -> str:
return (
self.__class__.__name__
+ '('
+ 'num_ang_bins='
+ str(self.num_ang_bins)
+ ', '
+ 'num_spatial_bins='
+ str(self.num_spatial_bins)
+ ', '
+ 'patch_size='
+ str(self.patch_size)
+ ', '
+ 'rootsift='
+ str(self.rootsift)
+ ', '
+ 'clipval='
+ str(self.clipval)
+ ')'
)
def __init__(
self,
patch_size: int = 41,
num_ang_bins: int = 8,
num_spatial_bins: int = 4,
rootsift: bool = True,
clipval: float = 0.2,
) -> None:
super().__init__()
self.eps = 1e-10
self.num_ang_bins = num_ang_bins
self.num_spatial_bins = num_spatial_bins
self.clipval = clipval
self.rootsift = rootsift
self.patch_size = patch_size
ks: int = self.patch_size
sigma: float = float(ks) / math.sqrt(2.0)
self.gk = get_gaussian_kernel2d((ks, ks), (sigma, sigma), True)
(self.bin_ksize, self.bin_stride, self.pad) = get_sift_bin_ksize_stride_pad(patch_size, num_spatial_bins)
nw = get_sift_pooling_kernel(ksize=self.bin_ksize).float()
self.pk = nn.Conv2d(
1,
1,
kernel_size=(nw.size(0), nw.size(1)),
stride=(self.bin_stride, self.bin_stride),
padding=(self.pad, self.pad),
bias=False,
)
self.pk.weight.data.copy_(nw.reshape(1, 1, nw.size(0), nw.size(1))) # type: ignore
return
def get_pooling_kernel(self) -> torch.Tensor:
return self.pk.weight.detach()
def get_weighting_kernel(self) -> torch.Tensor:
return self.gk.detach()
def forward(self, input):
if not isinstance(input, torch.Tensor):
raise TypeError(f"Input type is not a torch.Tensor. Got {type(input)}")
if not len(input.shape) == 4:
raise ValueError(f"Invalid input shape, we expect Bx1xHxW. Got: {input.shape}")
B, CH, W, H = input.size()
if (W != self.patch_size) or (H != self.patch_size) or (CH != 1):
raise TypeError(
"input shape should be must be [Bx1x{}x{}]. "
"Got {}".format(self.patch_size, self.patch_size, input.size())
)
self.pk = self.pk.to(input.dtype).to(input.device)
grads: torch.Tensor = spatial_gradient(input, 'diff')
# unpack the edges
gx: torch.Tensor = grads[:, :, 0]
gy: torch.Tensor = grads[:, :, 1]
mag: torch.Tensor = torch.sqrt(gx * gx + gy * gy + self.eps)
ori: torch.Tensor = torch.atan2(gy, gx + self.eps) + 2.0 * pi
mag = mag * self.gk.expand_as(mag).type_as(mag).to(mag.device)
o_big: torch.Tensor = float(self.num_ang_bins) * ori / (2.0 * pi)
bo0_big_: torch.Tensor = torch.floor(o_big)
wo1_big_: torch.Tensor = o_big - bo0_big_
bo0_big: torch.Tensor = bo0_big_ % self.num_ang_bins
bo1_big: torch.Tensor = (bo0_big + 1) % self.num_ang_bins
wo0_big: torch.Tensor = (1.0 - wo1_big_) * mag # type: ignore
wo1_big: torch.Tensor = wo1_big_ * mag
ang_bins = []
for i in range(0, self.num_ang_bins):
out = self.pk((bo0_big == i).to(input.dtype) * wo0_big + (bo1_big == i).to(input.dtype) * wo1_big)
ang_bins.append(out)
ang_bins = torch.cat(ang_bins, dim=1)
ang_bins = ang_bins.view(B, -1)
ang_bins = F.normalize(ang_bins, p=2)
ang_bins = torch.clamp(ang_bins, 0.0, float(self.clipval))
ang_bins = F.normalize(ang_bins, p=2)
if self.rootsift:
ang_bins = torch.sqrt(F.normalize(ang_bins, p=1) + self.eps)
return ang_bins
def sift_describe(
input: torch.Tensor,
patch_size: int = 41,
num_ang_bins: int = 8,
num_spatial_bins: int = 4,
rootsift: bool = True,
clipval: float = 0.2,
) -> torch.Tensor:
r"""Compute the sift descriptor.
See :class:`~kornia.feature.SIFTDescriptor` for details.
"""
return SIFTDescriptor(patch_size, num_ang_bins, num_spatial_bins, rootsift, clipval)(input)