|
r""" Conovlutional Hough matching layers """ |
|
|
|
import torch.nn as nn |
|
import torch |
|
|
|
from .base.correlation import Correlation |
|
from .base.geometry import Geometry |
|
from .base.chm import CHM4d, CHM6d |
|
|
|
|
|
class CHMLearner(nn.Module): |
|
|
|
def __init__(self, ktype, feat_dim): |
|
super(CHMLearner, self).__init__() |
|
|
|
|
|
self.scales = [0.5, 1, 2] |
|
self.conv2ds = nn.ModuleList([nn.Conv2d(feat_dim, feat_dim // 4, kernel_size=3, padding=1, bias=False) for _ in self.scales]) |
|
|
|
|
|
ksz_translation = 5 |
|
ksz_scale = 3 |
|
self.chm6d = CHM6d(1, 1, ksz_scale, ksz_translation, ktype) |
|
self.chm4d = CHM4d(1, 1, ksz_translation, ktype, bias=True) |
|
|
|
|
|
self.relu = nn.ReLU(inplace=True) |
|
self.sigmoid = nn.Sigmoid() |
|
self.softplus = nn.Softplus() |
|
|
|
def forward(self, src_feat, trg_feat): |
|
|
|
corr = Correlation.build_correlation6d(src_feat, trg_feat, self.scales, self.conv2ds).unsqueeze(1) |
|
bsz, ch, s, s, h, w, h, w = corr.size() |
|
|
|
|
|
corr = self.chm6d(corr) |
|
corr = self.sigmoid(corr) |
|
|
|
|
|
corr = corr.view(bsz, -1, h, w, h, w).max(dim=1)[0] |
|
corr = Geometry.interpolate4d(corr, [h * 2, w * 2]).unsqueeze(1) |
|
|
|
|
|
corr = self.chm4d(corr).squeeze(1) |
|
|
|
|
|
corr = self.softplus(corr) |
|
corr = Correlation.mutual_nn_filter(corr.view(bsz, corr.size(-1) ** 2, corr.size(-1) ** 2).contiguous()) |
|
|
|
return corr |
|
|
|
|