Skip to content

dncnn

Classes:

  • ConvBlock –

    Convolution block: conv => BN => act.

  • DnCNN –

    Implementation of the DnCNN architecture from [1].

ConvBlock 

ConvBlock(
    in_ch: int,
    out_ch: int,
    kernel_size: int,
    pad_mode: str = "replicate",
    last_block: bool = False,
)

Bases: Sequential

Convolution block: conv => BN => act.

Source code in src/autoden/models/dncnn.py 
 8
 9
10
11
12
13
14
15
16
17
def __init__(self, in_ch: int, out_ch: int, kernel_size: int, pad_mode: str = "replicate", last_block: bool = False):
    pad_size = (kernel_size - 1) // 2
    if last_block:
        post_conv = []
    else:
        post_conv = [nn.BatchNorm2d(out_ch), nn.LeakyReLU(0.2, inplace=True)]
    super().__init__(
        nn.Conv2d(in_ch, out_ch, kernel_size=kernel_size, padding=pad_size, padding_mode=pad_mode, bias=False),
        *post_conv,
    )

DnCNN 

DnCNN(
    n_channels_in: int,
    n_channels_out: int,
    n_layers: int = 20,
    n_features: int = 32,
    kernel_size: int = 3,
    pad_mode: str = "replicate",
    device: str = "cuda" if is_available() else "cpu",
)

Bases: Sequential

Implementation of the DnCNN architecture from [1].

[1] Zhang, et al., "Beyond a Gaussian denoiser: Residual learning of deep CNN for image denoising," IEEE Trans. on Image Processing, 2017.

Source code in src/autoden/models/dncnn.py 
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
def __init__(
    self,
    n_channels_in: int,
    n_channels_out: int,
    n_layers: int = 20,
    n_features: int = 32,
    kernel_size: int = 3,
    pad_mode: str = "replicate",
    device: str = "cuda" if pt.cuda.is_available() else "cpu",
):
    init_params = locals()
    del init_params["self"]
    del init_params["__class__"]
    # From zhang-2017-beyon-gauss-denois:
    #
    #  Thus, for Gaussian denoising with a certain noise level, we
    #  set the receptive field size of DnCNN to 35 × 35 with the
    #  corresponding depth of 17. For other general image denoising
    #  tasks, we adopt a larger receptive field and set the depth
    #  to be 20.
    #
    # Hence, we set the standard depth to 20.
    layers = [
        ConvBlock(
            n_channels_in if i_l == 0 else n_features,
            n_channels_out if i_l == (n_layers - 1) else n_features,
            kernel_size=kernel_size,
            pad_mode=pad_mode,
            last_block=(i_l == (n_layers - 1)),
        )
        for i_l in range(n_layers)
    ]

    super().__init__(*layers)
    self.init_params = init_params
    self.device = device

    self.to(self.device)