实验方法

1. Hyp-NeRF

if encoding == 'hashgrid':
from gridencoder import GridEncoder
encoder = GridEncoder(input_dim=input_dim, num_levels=num_levels, level_dim=level_dim, base_resolution=base_resolution, log2_hashmap_size=log2_hashmap_size, desired_resolution=desired_resolution, gridtype='hash', align_corners=align_corners, num_instances=num_instances)

2. R2L

3. GRAM-HD

## upsamping 上采样
self.num_upconvs = int(np.log2(scale_factor))

## patch-gan -->  discriminator()

class GramEncoderPatchDiscriminator(Discriminator):
    def __init__(self, img_size, img_channels=3, norm_layer=nn.BatchNorm2d):
        super().__init__()
        self.img_size = img_size
        self.img_size_log2 = int(np.log2(img_size))

        self.layers = nn.ModuleList([])
        for i in range(13 - self.img_size_log2, 12):
            self.layers.append(ResidualCoordConvBlock(int(min(400, 2**i)), int(min(400, 2**(i+1))), 
                                                      downsample=True))
        self.fromRGB = AdapterBlock(img_channels, int(min(400, 2**(13 - self.img_size_log2))))
        self.final_layer = nn.Conv2d(400, 1 + 2, 2)

        # patch-gan
        self.patchgan = NLayerDiscriminator(img_channels, min(64, int(2**(14-self.img_size_log2))), 
                                            max(3,self.img_size_log2-6), norm_layer=norm_layer)

    def forward(self, input):
        x = self.fromRGB(input)
        for layer in self.layers:
            x = layer(x)

        x = self.final_layer(x).reshape(x.shape[0], -1)

        prediction = x[..., 0:1]
        position = x[..., 1:]

        patch = self.patchgan(input).reshape(x.shape[0], -1)

        return prediction, position, patch

class NLayerDiscriminator(nn.Module):
    """Defines a PatchGAN discriminator"""

    def __init__(self, input_nc, ndf=64, n_layers=3, norm_layer=nn.BatchNorm2d):
        """Construct a PatchGAN discriminator
        Parameters:
            input_nc (int)  -- the number of channels in input images
            ndf (int)       -- the number of filters in the last conv layer
            n_layers (int)  -- the number of conv layers in the discriminator
            norm_layer      -- normalization layer
        """
        super(NLayerDiscriminator, self).__init__()
        if type(norm_layer) == functools.partial:  # no need to use bias as BatchNorm2d has affine parameters
            use_bias = norm_layer.func == nn.InstanceNorm2d
        else:
            use_bias = norm_layer == nn.InstanceNorm2d

        kw = 4
        padw = 1
        sequence = [nn.Conv2d(input_nc, ndf, kernel_size=kw, stride=2, padding=padw), nn.LeakyReLU(0.2, True)]
        nf_mult = 1
        nf_mult_prev = 1
        for n in range(1, n_layers):  # gradually increase the number of filters
            nf_mult_prev = nf_mult
            nf_mult = min(2 ** n, 8)
            sequence += [
                nn.Conv2d(ndf * nf_mult_prev, ndf * nf_mult, kernel_size=kw, stride=2, padding=padw, 
                          bias=use_bias),
                norm_layer(ndf * nf_mult),
                nn.LeakyReLU(0.2, True)
            ]

        nf_mult_prev = nf_mult
        nf_mult = min(2 ** n_layers, 8)
        sequence += [
            nn.Conv2d(ndf * nf_mult_prev, ndf * nf_mult, kernel_size=kw, stride=1, padding=padw, bias=use_bias),
            norm_layer(ndf * nf_mult),
            nn.LeakyReLU(0.2, True)
        ]
		
        # output 1 channel prediction map
        sequence += [nn.Conv2d(ndf * nf_mult, 1, kernel_size=kw, stride=1, padding=padw)]  
        self.model = nn.Sequential(*sequence)

    def forward(self, input):
        """Standard forward."""
        return self.model(input)