"""
CKDN model introduced by
Zheng, Heliang, Huan Yang, Jianlong Fu, Zheng-Jun Zha, and Jiebo Luo.
"Learning conditional knowledge distillation for degraded-reference image quality assessment."
In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pp. 10242-10251. 2021.
Ref url: https://github.com/researchmm/CKDN.
Re-implemented by: Chaofeng Chen (https://github.com/chaofengc)
"""
import torch
import torch.nn as nn
import math
import torchvision as tv
from pyiqa.utils.registry import ARCH_REGISTRY
from pyiqa.archs.arch_util import load_pretrained_network
try:
from torch.hub import load_state_dict_from_url
except ImportError:
from torch.utils.model_zoo import load_url as load_state_dict_from_url
from pyiqa.archs.arch_util import get_url_from_name
[docs]
default_model_urls = {'url': get_url_from_name('CKDN_model_best-38b27dc6.pth')}
[docs]
model_urls = {
'resnet50': 'https://download.pytorch.org/models/resnet50-19c8e357.pth',
}
[docs]
def conv3x3(in_planes, out_planes, stride=1, groups=1, dilation=1):
"""3x3 convolution with padding"""
return nn.Conv2d(
in_planes,
out_planes,
kernel_size=3,
stride=stride,
padding=dilation,
groups=groups,
bias=False,
dilation=dilation,
)
[docs]
def conv1x1(in_planes, out_planes, stride=1):
"""1x1 convolution"""
return nn.Conv2d(in_planes, out_planes, kernel_size=1, stride=stride, bias=False)
[docs]
class BasicBlock(nn.Module):
expansion = 1
__constants__ = ['downsample']
def __init__(
self,
inplanes,
planes,
stride=1,
downsample=None,
groups=1,
base_width=64,
dilation=1,
norm_layer=None,
):
super(BasicBlock, self).__init__()
if norm_layer is None:
norm_layer = nn.BatchNorm2d
if groups != 1 or base_width != 64:
raise ValueError('BasicBlock only supports groups=1 and base_width=64')
if dilation > 1:
raise NotImplementedError('Dilation > 1 not supported in BasicBlock')
self.conv1 = conv3x3(inplanes, planes, stride)
self.bn1 = norm_layer(planes)
self.relu = nn.ReLU(inplace=True)
self.conv2 = conv3x3(planes, planes)
self.bn2 = norm_layer(planes)
self.downsample = downsample
self.stride = stride
[docs]
def forward(self, x):
identity = x
out = self.conv1(x)
out = self.bn1(out)
out = self.relu(out)
out = self.conv2(out)
out = self.bn2(out)
if self.downsample is not None:
identity = self.downsample(x)
out += identity
out = self.relu(out)
return out
[docs]
class Bottleneck(nn.Module):
expansion = 4
__constants__ = ['downsample']
def __init__(
self,
inplanes,
planes,
stride=1,
downsample=None,
groups=1,
base_width=64,
dilation=1,
norm_layer=None,
):
super(Bottleneck, self).__init__()
if norm_layer is None:
norm_layer = nn.BatchNorm2d
width = int(planes * (base_width / 64.0)) * groups
self.conv1 = conv1x1(inplanes, width)
self.bn1 = norm_layer(width)
self.conv2 = conv3x3(width, width, stride, groups, dilation)
self.bn2 = norm_layer(width)
self.conv3 = conv1x1(width, planes * self.expansion)
self.bn3 = norm_layer(planes * self.expansion)
self.relu = nn.ReLU(inplace=True)
self.downsample = downsample
self.stride = stride
[docs]
def forward(self, x):
identity = x
out = self.conv1(x)
out = self.bn1(out)
out = self.relu(out)
out = self.conv2(out)
out = self.bn2(out)
out = self.relu(out)
out = self.conv3(out)
out = self.bn3(out)
if self.downsample is not None:
identity = self.downsample(x)
out += identity
out = self.relu(out)
return out
[docs]
class ResNet(nn.Module):
def __init__(
self,
block,
layers,
num_classes=1000,
zero_init_residual=False,
groups=1,
width_per_group=64,
replace_stride_with_dilation=None,
norm_layer=None,
):
super(ResNet, self).__init__()
if norm_layer is None:
norm_layer = nn.BatchNorm2d
self._norm_layer = norm_layer
self.inplanes = 64
self.dilation = 1
self.k = 3
if replace_stride_with_dilation is None:
replace_stride_with_dilation = [False, False, False]
if len(replace_stride_with_dilation) != 3:
raise ValueError(
'replace_stride_with_dilation should be None or a 3-element tuple, got {}'.format(
replace_stride_with_dilation
)
)
self.groups = groups
self.base_width = width_per_group
self.head = 8
self.qse_1 = nn.Conv2d(
3, self.inplanes, kernel_size=7, stride=2, padding=3, bias=False
)
self.qse_2 = self._make_layer(block, 64, layers[0])
self.csp = self._make_layer(block, 128, layers[1], stride=2, dilate=False)
self.inplanes = 64
self.dte_1 = nn.Conv2d(
3, self.inplanes, kernel_size=7, stride=2, padding=3, bias=False
)
self.dte_2 = self._make_layer(block, 64, layers[0])
self.aux_csp = self._make_layer(block, 128, layers[1], stride=2, dilate=False)
self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1)
self.avgpool = nn.AdaptiveAvgPool2d((1, 1))
self.fc_ = nn.Sequential(
nn.Linear((512) * 1 * 1, 2048),
nn.ReLU(True),
nn.Dropout(),
nn.Linear(2048, 2048),
nn.ReLU(True),
nn.Dropout(),
nn.Linear(2048, 1),
)
self.fc1_ = nn.Sequential(
nn.Linear((512) * 1 * 1, 2048),
nn.ReLU(True),
nn.Dropout(),
nn.Linear(2048, 2048),
nn.ReLU(True),
nn.Dropout(),
nn.Linear(2048, 1),
)
for m in self.modules():
if isinstance(m, nn.Conv2d):
nn.init.kaiming_normal_(m.weight, mode='fan_out', nonlinearity='relu')
elif isinstance(m, (nn.BatchNorm2d, nn.GroupNorm)):
nn.init.constant_(m.weight, 1)
nn.init.constant_(m.bias, 0)
if zero_init_residual:
for m in self.modules():
if isinstance(m, Bottleneck):
nn.init.constant_(m.bn3.weight, 0)
elif isinstance(m, BasicBlock):
nn.init.constant_(m.bn2.weight, 0)
def _make_layer(self, block, planes, blocks, stride=1, dilate=False):
norm_layer = self._norm_layer
downsample = None
previous_dilation = self.dilation
if dilate:
self.dilation *= stride
stride = 1
if stride != 1 or self.inplanes != planes * block.expansion:
downsample = nn.Sequential(
conv1x1(self.inplanes, planes * block.expansion, stride),
norm_layer(planes * block.expansion),
)
layers = []
layers.append(
block(
self.inplanes,
planes,
stride,
downsample,
self.groups,
self.base_width,
previous_dilation,
norm_layer,
)
)
self.inplanes = planes * block.expansion
for _ in range(1, blocks):
layers.append(
block(
self.inplanes,
planes,
groups=self.groups,
base_width=self.base_width,
dilation=self.dilation,
norm_layer=norm_layer,
)
)
return nn.Sequential(*layers)
[docs]
def forward(self, x, y):
"""
Forward pass for the ResNet model.
Args:
x (torch.Tensor): Input tensor with shape (N, C, H, W).
y (torch.Tensor): Reference tensor with shape (N, C, H, W).
Returns:
torch.Tensor: Output tensor after processing through the network.
"""
rest1 = x
dist1 = y
rest1 = self.qse_2(self.maxpool(self.qse_1(rest1)))
dist1 = self.dte_2(self.maxpool(self.dte_1(dist1)))
x = rest1 - dist1
x = self.csp(x)
x = self.avgpool(x)
x = torch.flatten(x, 1)
dr = torch.sigmoid(self.fc_(x))
return dr
def _resnet(arch, block, layers, pretrained, progress, **kwargs):
"""
Helper function to create a ResNet model.
Args:
arch (str): Architecture name.
block (nn.Module): Block type (BasicBlock or Bottleneck).
layers (list): List of layer configurations.
pretrained (bool): Whether to load pretrained weights.
progress (bool): Whether to display progress bar.
**kwargs: Additional arguments.
Returns:
ResNet: Instantiated ResNet model.
"""
model = ResNet(block, layers, **kwargs)
if pretrained:
state_dict = load_state_dict_from_url(model_urls[arch], progress=progress)
keys = state_dict.keys()
for key in list(keys):
if 'conv1' in key:
state_dict[key.replace('conv1', 'qse_1')] = state_dict[key]
state_dict[key.replace('conv1', 'dte_1')] = state_dict[key]
if 'layer1' in key:
state_dict[key.replace('layer1', 'qse_2')] = state_dict[key]
state_dict[key.replace('layer1', 'dte_2')] = state_dict[key]
if 'layer2' in key:
state_dict[key.replace('layer2', 'csp')] = state_dict[key]
state_dict[key.replace('layer2', 'aux_csp')] = state_dict[key]
model.load_state_dict(state_dict, strict=False)
return model
@ARCH_REGISTRY.register()
[docs]
class CKDN(nn.Module):
"""
CKDN metric.
Args:
pretrained_model_path (str): The model path.
use_default_preprocess (bool): Whether to use default preprocess, default: True.
default_mean (tuple): The mean value.
default_std (tuple): The std value.
Reference:
Zheng, Heliang, Huan Yang, Jianlong Fu, Zheng-Jun Zha, and Jiebo Luo.
"Learning conditional knowledge distillation for degraded-reference image
quality assessment." In Proceedings of the IEEE/CVF International Conference
on Computer Vision (ICCV), pp. 10242-10251. 2021.
"""
def __init__(
self,
pretrained=True,
pretrained_model_path=None,
use_default_preprocess=True,
default_mean=(0.485, 0.456, 0.406),
default_std=(0.229, 0.224, 0.225),
**kwargs,
):
super().__init__()
self.net = _resnet('resnet50', Bottleneck, [3, 4, 6, 3], True, True, **kwargs)
self.use_default_preprocess = use_default_preprocess
self.default_mean = torch.Tensor(default_mean).view(1, 3, 1, 1)
self.default_std = torch.Tensor(default_std).view(1, 3, 1, 1)
if pretrained_model_path is not None:
load_pretrained_network(self, pretrained_model_path)
elif pretrained:
load_pretrained_network(self, default_model_urls['url'])
def _default_preprocess(self, x, y):
"""
Default preprocessing of CKDN.
Args:
x (torch.Tensor): Input tensor with shape (N, C, H, W) in RGB format; value range, 0 ~ 1.
y (torch.Tensor): Reference tensor with shape (N, C, H, W) in RGB format; value range, 0 ~ 1.
Returns:
tuple: Preprocessed tensors (x, y).
"""
scaled_size = int(math.floor(288 / 0.875))
x = tv.transforms.functional.resize(
x, scaled_size, tv.transforms.InterpolationMode.BICUBIC
)
y = tv.transforms.functional.resize(
y, scaled_size, tv.transforms.InterpolationMode.NEAREST
)
x = tv.transforms.functional.center_crop(x, 288)
y = tv.transforms.functional.center_crop(y, 288)
x = (x - self.default_mean.to(x)) / self.default_std.to(x)
y = (y - self.default_mean.to(y)) / self.default_std.to(y)
return x, y
[docs]
def forward(self, x, y):
"""
Compute IQA using CKDN model.
Args:
x (torch.Tensor): Input tensor with shape (N, C, H, W). RGB channel order for colour images.
y (torch.Tensor): Reference tensor with shape (N, C, H, W). RGB channel order for colour images.
Returns:
torch.Tensor: Value of CKDN model.
"""
if self.use_default_preprocess:
x, y = self._default_preprocess(x, y)
return self.net(x, y)