BANet.py

""" BANet
Paper: ``
    - https://arxiv.org/abs/2104.01136

ResT code and weights: https://github.com/wofmanaf/ResT
"""
import torch
import torch.nn as nn
from torch.nn import Module, Conv2d, Parameter, Softmax
from torchvision import models
from timm.models.layers import DropPath, to_2tuple, trunc_normal_


class Mlp(nn.Module):
    def __init__(self, in_features, hidden_features=None, out_features=None, act_layer=nn.GELU, drop=0.):
        super().__init__()
        out_features = out_features or in_features
        hidden_features = hidden_features or in_features
        self.fc1 = nn.Linear(in_features, hidden_features)
        self.act = act_layer()
        self.fc2 = nn.Linear(hidden_features, out_features)
        self.drop = nn.Dropout(drop)

    def forward(self, x):
        x = self.fc1(x)
        x = self.act(x)
        x = self.drop(x)
        x = self.fc2(x)
        x = self.drop(x)
        return x


class Attention(nn.Module):
    def __init__(self,
                 dim,
                 num_heads=8,
                 qkv_bias=False,
                 qk_scale=None,
                 attn_drop=0.,
                 proj_drop=0.,
                 sr_ratio=1,
                 apply_transform=False):
        super().__init__()
        self.num_heads = num_heads
        head_dim = dim // num_heads
        self.scale = qk_scale or head_dim ** -0.5

        self.q = nn.Linear(dim, dim, bias=qkv_bias)
        self.kv = nn.Linear(dim, dim * 2, bias=qkv_bias)
        self.attn_drop = nn.Dropout(attn_drop)
        self.proj = nn.Linear(dim, dim)
        self.proj_drop = nn.Dropout(proj_drop)

        self.sr_ratio = sr_ratio
        if sr_ratio > 1:
            self.sr = nn.Conv2d(dim, dim, kernel_size=sr_ratio+1, stride=sr_ratio, padding=sr_ratio // 2, groups=dim)
            self.sr_norm = nn.LayerNorm(dim)

        self.apply_transform = apply_transform and num_heads > 1
        if self.apply_transform:
            self.transform_conv = nn.Conv2d(self.num_heads, self.num_heads, kernel_size=1, stride=1)
            self.transform_norm = nn.InstanceNorm2d(self.num_heads)

    def forward(self, x, H, W):
        B, N, C = x.shape
        q = self.q(x).reshape(B, N, self.num_heads, C // self.num_heads).permute(0, 2, 1, 3)
        if self.sr_ratio > 1:
            x_ = x.permute(0, 2, 1).reshape(B, C, H, W)
            x_ = self.sr(x_).reshape(B, C, -1).permute(0, 2, 1)
            x_ = self.sr_norm(x_)
            kv = self.kv(x_).reshape(B, -1, 2, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4)
        else:
            kv = self.kv(x).reshape(B, N, 2, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4)
        k, v = kv[0], kv[1]

        attn = (q @ k.transpose(-2, -1)) * self.scale
        if self.apply_transform:
            attn = self.transform_conv(attn)
            attn = attn.softmax(dim=-1)
            attn = self.transform_norm(attn)
        else:
            attn = attn.softmax(dim=-1)

        attn = self.attn_drop(attn)
        x = (attn @ v).transpose(1, 2).reshape(B, N, C)
        x = self.proj(x)
        x = self.proj_drop(x)
        return x


class Block(nn.Module):
    def __init__(self, dim, num_heads, mlp_ratio=4., qkv_bias=False, qk_scale=None, drop=0., attn_drop=0.,
                 drop_path=0., act_layer=nn.GELU, norm_layer=nn.LayerNorm, sr_ratio=1, apply_transform=False):
        super().__init__()
        self.norm1 = norm_layer(dim)
        self.attn = Attention(
            dim, num_heads=num_heads, qkv_bias=qkv_bias, qk_scale=qk_scale,
            attn_drop=attn_drop, proj_drop=drop, sr_ratio=sr_ratio, apply_transform=apply_transform)
        # NOTE: drop path for stochastic depth, we shall see if this is better than dropout here
        self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity()
        self.norm2 = norm_layer(dim)
        mlp_hidden_dim = int(dim * mlp_ratio)
        self.mlp = Mlp(in_features=dim, hidden_features=mlp_hidden_dim, act_layer=act_layer, drop=drop)

    def forward(self, x, H, W):
        x = x + self.drop_path(self.attn(self.norm1(x), H, W))
        x = x + self.drop_path(self.mlp(self.norm2(x)))
        return x


class PA(nn.Module):
    def __init__(self, dim):
        super().__init__()
        self.pa_conv = nn.Conv2d(dim, dim, kernel_size=3, padding=1, groups=dim)
        self.sigmoid = nn.Sigmoid()

    def forward(self, x):
        return x * self.sigmoid(self.pa_conv(x))


class GL(nn.Module):
    def __init__(self, dim):
        super().__init__()
        self.gl_conv = nn.Conv2d(dim, dim, kernel_size=3, padding=1, groups=dim)

    def forward(self, x):
        return x + self.gl_conv(x)


class PatchEmbed(nn.Module):
    """ Image to Patch Embedding"""
    def __init__(self, patch_size=16, in_ch=3, out_ch=768, with_pos=False):
        super().__init__()
        self.patch_size = to_2tuple(patch_size)
        self.conv = nn.Conv2d(in_ch, out_ch, kernel_size=patch_size+1, stride=patch_size, padding=patch_size // 2)
        self.norm = nn.BatchNorm2d(out_ch)

        self.with_pos = with_pos
        if self.with_pos:
            self.pos = PA(out_ch)

    def forward(self, x):
        B, C, H, W = x.shape
        x = self.conv(x)
        x = self.norm(x)
        if self.with_pos:
            x = self.pos(x)
        x = x.flatten(2).transpose(1, 2)
        H, W = H // self.patch_size[0], W // self.patch_size[1]
        return x, (H, W)


class BasicStem(nn.Module):
    def __init__(self, in_ch=3, out_ch=64, with_pos=False):
        super(BasicStem, self).__init__()
        hidden_ch = out_ch // 2
        self.conv1 = nn.Conv2d(in_ch, hidden_ch, kernel_size=3, stride=2, padding=1, bias=False)
        self.norm1 = nn.BatchNorm2d(hidden_ch)
        self.conv2 = nn.Conv2d(hidden_ch, hidden_ch, kernel_size=3, stride=1, padding=1, bias=False)
        self.norm2 = nn.BatchNorm2d(hidden_ch)
        self.conv3 = nn.Conv2d(hidden_ch, out_ch, kernel_size=3, stride=2, padding=1, bias=False)

        self.act = nn.ReLU(inplace=True)
        self.with_pos = with_pos
        if self.with_pos:
            self.pos = PA(out_ch)

    def forward(self, x):
        x = self.conv1(x)
        x = self.norm1(x)
        x = self.act(x)

        x = self.conv2(x)
        x = self.norm2(x)
        x = self.act(x)

        x = self.conv3(x)
        if self.with_pos:
            x = self.pos(x)
        return x


class ResT(nn.Module):
    def __init__(self, in_chans=3, num_classes=1000, embed_dims=[64, 128, 256, 512],
                 num_heads=[1, 2, 4, 8], mlp_ratios=[4, 4, 4, 4], qkv_bias=False,
                 qk_scale=None, drop_rate=0., attn_drop_rate=0., drop_path_rate=0.,
                 depths=[2, 2, 2, 2], sr_ratios=[8, 4, 2, 1],
                 norm_layer=nn.LayerNorm, apply_transform=False):
        super().__init__()
        self.num_classes = num_classes
        self.depths = depths
        self.apply_transform = apply_transform

        self.stem = BasicStem(in_ch=in_chans, out_ch=embed_dims[0], with_pos=True)

        self.patch_embed_2 = PatchEmbed(patch_size=2, in_ch=embed_dims[0], out_ch=embed_dims[1], with_pos=True)
        self.patch_embed_3 = PatchEmbed(patch_size=2, in_ch=embed_dims[1], out_ch=embed_dims[2], with_pos=True)
        self.patch_embed_4 = PatchEmbed(patch_size=2, in_ch=embed_dims[2], out_ch=embed_dims[3], with_pos=True)

        # transformer encoder
        dpr = [x.item() for x in torch.linspace(0, drop_path_rate, sum(depths))]
        cur = 0

        self.stage1 = nn.ModuleList([
            Block(embed_dims[0], num_heads[0], mlp_ratios[0], qkv_bias, qk_scale, drop_rate, attn_drop_rate,
                  drop_path=dpr[cur+i], norm_layer=norm_layer, sr_ratio=sr_ratios[0], apply_transform=apply_transform)
            for i in range(self.depths[0])])

        cur += depths[0]
        self.stage2 = nn.ModuleList([
            Block(embed_dims[1], num_heads[1], mlp_ratios[1], qkv_bias, qk_scale, drop_rate, attn_drop_rate,
                  drop_path=dpr[cur+i], norm_layer=norm_layer, sr_ratio=sr_ratios[1], apply_transform=apply_transform)
            for i in range(self.depths[1])])

        cur += depths[1]
        self.stage3 = nn.ModuleList([
            Block(embed_dims[2], num_heads[2], mlp_ratios[2], qkv_bias, qk_scale, drop_rate, attn_drop_rate,
                  drop_path=dpr[cur+i], norm_layer=norm_layer, sr_ratio=sr_ratios[2], apply_transform=apply_transform)
            for i in range(self.depths[2])])

        cur += depths[2]
        self.stage4 = nn.ModuleList([
            Block(embed_dims[3], num_heads[3], mlp_ratios[3], qkv_bias, qk_scale, drop_rate, attn_drop_rate,
                  drop_path=dpr[cur+i], norm_layer=norm_layer, sr_ratio=sr_ratios[3], apply_transform=apply_transform)
            for i in range(self.depths[3])])

        self.norm = norm_layer(embed_dims[3])

        # init weights
        self.apply(self._init_weights)

    def _init_weights(self, m):
        if isinstance(m, nn.Conv2d):
            trunc_normal_(m.weight, std=0.02)
        elif isinstance(m, nn.Linear):
            trunc_normal_(m.weight, std=0.02)
            if m.bias is not None:
                nn.init.constant_(m.bias, 0)
        elif isinstance(m, (nn.LayerNorm, nn.BatchNorm2d)):
            nn.init.constant_(m.weight, 1.0)
            nn.init.constant_(m.bias, 0)

    def forward(self, x):
        x = self.stem(x)
        B, _, H, W = x.shape
        x = x.flatten(2).permute(0, 2, 1)

        # stage 1
        for blk in self.stage1:
            x = blk(x, H, W)
        x = x.permute(0, 2, 1).reshape(B, -1, H, W)
        # x1 = x

        # stage 2
        x, (H, W) = self.patch_embed_2(x)
        for blk in self.stage2:
            x = blk(x, H, W)
        x = x.permute(0, 2, 1).reshape(B, -1, H, W)
        # x2 = x

        # stage 3
        x, (H, W) = self.patch_embed_3(x)
        for blk in self.stage3:
            x = blk(x, H, W)
        x = x.permute(0, 2, 1).reshape(B, -1, H, W)
        x3 = x

        # stage 4
        x, (H, W) = self.patch_embed_4(x)
        for blk in self.stage4:
            x = blk(x, H, W)
        x = self.norm(x)
        x = x.permute(0, 2, 1).reshape(B, -1, H, W)
        x4 = x

        return x3, x4


def rest_lite(pretrained=True, weight_path='pretrain_weights/rest_lite.pth',  **kwargs):
    model = ResT(embed_dims=[64, 128, 256, 512], num_heads=[1, 2, 4, 8], mlp_ratios=[4, 4, 4, 4], qkv_bias=True,
                 depths=[2, 2, 2, 2], sr_ratios=[8, 4, 2, 1], apply_transform=True, **kwargs)
    if pretrained and weight_path is not None:
        old_dict = torch.load(weight_path)
        model_dict = model.state_dict()
        old_dict = {k: v for k, v in old_dict.items() if (k in model_dict)}
        model_dict.update(old_dict)
        model.load_state_dict(model_dict)
    return model


class ConvBNReLU(nn.Module):
    def __init__(self, in_chan, out_chan, ks=3, stride=1, padding=1):
        super(ConvBNReLU, self).__init__()
        self.conv = nn.Conv2d(in_chan,
                              out_chan,
                              kernel_size=ks,
                              stride=stride,
                              padding=padding,
                              bias=False)
        self.bn = nn.BatchNorm2d(out_chan)
        self.relu = nn.ReLU(inplace=True)
        self.init_weight()

    def forward(self, x):
        x = self.conv(x)
        x = self.bn(x)
        x = self.relu(x)
        return x

    def init_weight(self):
        for ly in self.children():
            if isinstance(ly, nn.Conv2d):
                nn.init.kaiming_normal_(ly.weight, a=1)
                if not ly.bias is None: nn.init.constant_(ly.bias, 0)


def l2_norm(x):
    return torch.einsum("bcn, bn->bcn", x, 1 / torch.norm(x, p=2, dim=-2))


class LinearAttention(Module):
    def __init__(self, in_places, scale=8, eps=1e-6):
        super(LinearAttention, self).__init__()
        self.gamma = Parameter(torch.zeros(1))
        self.in_places = in_places
        self.l2_norm = l2_norm
        self.eps = eps

        self.query_conv = Conv2d(in_channels=in_places, out_channels=in_places // scale, kernel_size=1)
        self.key_conv = Conv2d(in_channels=in_places, out_channels=in_places // scale, kernel_size=1)
        self.value_conv = Conv2d(in_channels=in_places, out_channels=in_places, kernel_size=1)

    def forward(self, x):
        # Apply the feature map to the queries and keys
        batch_size, chnnels, width, height = x.shape
        Q = self.query_conv(x).view(batch_size, -1, width * height)
        K = self.key_conv(x).view(batch_size, -1, width * height)
        V = self.value_conv(x).view(batch_size, -1, width * height)

        Q = self.l2_norm(Q).permute(-3, -1, -2)
        K = self.l2_norm(K)

        tailor_sum = 1 / (width * height + torch.einsum("bnc, bc->bn", Q, torch.sum(K, dim=-1) + self.eps))
        value_sum = torch.einsum("bcn->bc", V).unsqueeze(-1)
        value_sum = value_sum.expand(-1, chnnels, width * height)

        matrix = torch.einsum('bmn, bcn->bmc', K, V)
        matrix_sum = value_sum + torch.einsum("bnm, bmc->bcn", Q, matrix)

        weight_value = torch.einsum("bcn, bn->bcn", matrix_sum, tailor_sum)
        weight_value = weight_value.view(batch_size, chnnels, height, width)

        return x + (self.gamma * weight_value).contiguous()


class Output(nn.Module):
    def __init__(self, in_chan, mid_chan, n_classes, up_factor=32, *args, **kwargs):
        super(Output, self).__init__()
        self.up_factor = up_factor
        out_chan = n_classes * up_factor * up_factor
        self.conv = ConvBNReLU(in_chan, mid_chan, ks=3, stride=1, padding=1)
        self.conv_out = nn.Conv2d(mid_chan, out_chan, kernel_size=1, bias=True)
        self.up = nn.PixelShuffle(up_factor)
        self.init_weight()

    def forward(self, x):
        x = self.conv(x)
        x = self.conv_out(x)
        x = self.up(x)
        return x

    def init_weight(self):
        for ly in self.children():
            if isinstance(ly, nn.Conv2d):
                nn.init.kaiming_normal_(ly.weight, a=1)
                if not ly.bias is None: nn.init.constant_(ly.bias, 0)

    def get_params(self):
        wd_params, nowd_params = [], []
        for name, module in self.named_modules():
            if isinstance(module, (nn.Linear, nn.Conv2d)):
                wd_params.append(module.weight)
                if not module.bias is None:
                    nowd_params.append(module.bias)
            elif isinstance(module, nn.modules.batchnorm._BatchNorm):
                nowd_params += list(module.parameters())
        return wd_params, nowd_params


class UpSample(nn.Module):

    def __init__(self, n_chan, factor=2):
        super(UpSample, self).__init__()
        out_chan = n_chan * factor * factor
        self.proj = nn.Conv2d(n_chan, out_chan, 1, 1, 0)
        self.up = nn.PixelShuffle(factor)
        self.init_weight()

    def forward(self, x):
        feat = self.proj(x)
        feat = self.up(feat)
        return feat

    def init_weight(self):
        nn.init.xavier_normal_(self.proj.weight, gain=1.)


class Attention_Embedding(nn.Module):
    def __init__(self, in_channels, out_channels):
        super(Attention_Embedding, self).__init__()
        self.attention = LinearAttention(in_channels)
        self.conv_attn = ConvBNReLU(in_channels, out_channels)
        self.up = UpSample(out_channels)

    def forward(self, high_feat, low_feat):
        A = self.attention(high_feat)
        A = self.conv_attn(A)
        A = self.up(A)

        output = low_feat * A
        output += low_feat

        return output


class FeatureAggregationModule(nn.Module):
    def __init__(self, in_chan, out_chan):
        super(FeatureAggregationModule, self).__init__()
        self.convblk = ConvBNReLU(in_chan, out_chan, ks=1, stride=1, padding=0)
        self.conv_atten = LinearAttention(out_chan)
        self.init_weight()

    def forward(self, fsp, fcp):
        fcat = torch.cat([fsp, fcp], dim=1)
        feat = self.convblk(fcat)
        atten = self.conv_atten(feat)
        feat_atten = torch.mul(feat, atten)
        feat_out = feat_atten + feat
        return feat_out

    def init_weight(self):
        for ly in self.children():
            if isinstance(ly, nn.Conv2d):
                nn.init.kaiming_normal_(ly.weight, a=1)
                if not ly.bias is None: nn.init.constant_(ly.bias, 0)

    def get_params(self):
        wd_params, nowd_params = [], []
        for name, module in self.named_modules():
            if isinstance(module, (nn.Linear, nn.Conv2d)):
                wd_params.append(module.weight)
                if not module.bias is None:
                    nowd_params.append(module.bias)
            elif isinstance(module, nn.modules.batchnorm._BatchNorm):
                nowd_params += list(module.parameters())
        return wd_params, nowd_params


class TexturePath(nn.Module):
    def __init__(self):
        super(TexturePath, self).__init__()
        self.conv1 = ConvBNReLU(3, 64, ks=7, stride=2, padding=3)
        self.conv2 = ConvBNReLU(64, 64, ks=3, stride=2, padding=1)
        self.conv3 = ConvBNReLU(64, 64, ks=3, stride=2, padding=1)
        self.conv_out = ConvBNReLU(64, 128, ks=1, stride=1, padding=0)
        self.init_weight()

    def forward(self, x):
        feat = self.conv1(x)
        feat = self.conv2(feat)
        feat = self.conv3(feat)
        feat = self.conv_out(feat)
        return feat

    def init_weight(self):
        for ly in self.children():
            if isinstance(ly, nn.Conv2d):
                nn.init.kaiming_normal_(ly.weight, a=1)
                if not ly.bias is None: nn.init.constant_(ly.bias, 0)

    def get_params(self):
        wd_params, nowd_params = [], []
        for name, module in self.named_modules():
            if isinstance(module, nn.Linear) or isinstance(module, nn.Conv2d):
                wd_params.append(module.weight)
                if not module.bias is None:
                    nowd_params.append(module.bias)
            elif isinstance(module, nn.modules.batchnorm._BatchNorm):
                nowd_params += list(module.parameters())
        return wd_params, nowd_params


class DependencyPath(nn.Module):
    def __init__(self, weight_path='pretrain_weights/rest_lite.pth'):
        super(DependencyPath, self).__init__()
        self.ResT = rest_lite(weight_path=weight_path)
        self.AE = Attention_Embedding(512, 256)
        self.conv_avg = ConvBNReLU(256, 128, ks=1, stride=1, padding=0)
        self.up = nn.Upsample(scale_factor=2.)

    def forward(self, x):
        e3, e4 = self.ResT(x)

        e = self.conv_avg(self.AE(e4, e3))

        return self.up(e)

    def get_params(self):
        wd_params, nowd_params = [], []
        for name, module in self.named_modules():
            if isinstance(module, (nn.Linear, nn.Conv2d)):
                wd_params.append(module.weight)
                if not module.bias is None:
                    nowd_params.append(module.bias)
            elif isinstance(module, nn.modules.batchnorm._BatchNorm):
                nowd_params += list(module.parameters())
        return wd_params, nowd_params


class DependencyPathRes(nn.Module):
    def __init__(self):
        super(DependencyPathRes, self).__init__()
        resnet = models.resnet18(True)
        self.firstconv = resnet.conv1
        self.firstbn = resnet.bn1
        self.firstrelu = resnet.relu
        self.firstmaxpool = resnet.maxpool
        self.encoder1 = resnet.layer1
        self.encoder2 = resnet.layer2
        self.encoder3 = resnet.layer3
        self.encoder4 = resnet.layer4
        self.AE = Attention_Embedding(512, 256)
        self.conv_avg = ConvBNReLU(256, 128, ks=1, stride=1, padding=0)
        self.up = nn.Upsample(scale_factor=2.)

    def forward(self, x):
        x1 = self.firstconv(x)
        x1 = self.firstbn(x1)
        x1 = self.firstrelu(x1)
        x1 = self.firstmaxpool(x1)
        e1 = self.encoder1(x1)
        e2 = self.encoder2(e1)
        e3 = self.encoder3(e2)
        e4 = self.encoder4(e3)

        e = self.conv_avg(self.AE(e4, e3))

        return self.up(e)

    def get_params(self):
        wd_params, nowd_params = [], []
        for name, module in self.named_modules():
            if isinstance(module, (nn.Linear, nn.Conv2d)):
                wd_params.append(module.weight)
                if not module.bias is None:
                    nowd_params.append(module.bias)
            elif isinstance(module, nn.modules.batchnorm._BatchNorm):
                nowd_params += list(module.parameters())
        return wd_params, nowd_params


class BANet(nn.Module):
    def __init__(self, num_classes=6, weight_path='pretrain_weights/rest_lite.pth'):
        super(BANet, self).__init__()    # path of pretrained weight file of ResT-lite  or None, recommend use.
        self.name = 'BANet'

        self.cp = DependencyPath(weight_path=weight_path)
        self.sp = TexturePath()
        self.fam = FeatureAggregationModule(256, 256)
        self.conv_out = Output(256, 256, num_classes, up_factor=8)
        self.init_weight()

    def forward(self, x):
        feat = self.cp(x)
        feat_sp = self.sp(x)
        feat_fuse = self.fam(feat_sp, feat)

        feat_out = self.conv_out(feat_fuse)

        return feat_out

    def init_weight(self):
        for ly in self.children():
            if isinstance(ly, nn.Conv2d):
                nn.init.kaiming_normal_(ly.weight, a=1)
                if not ly.bias is None: nn.init.constant_(ly.bias, 0)

    def get_params(self):
        wd_params, nowd_params, lr_mul_wd_params, lr_mul_nowd_params = [], [], [], []
        for name, child in self.named_children():
            child_wd_params, child_nowd_params = child.get_params()
            if isinstance(child, (Attention_Embedding, Output)):
                lr_mul_wd_params += child_wd_params
                lr_mul_nowd_params += child_nowd_params
            else:
                wd_params += child_wd_params
                nowd_params += child_nowd_params
        return wd_params, nowd_params, lr_mul_wd_params, lr_mul_nowd_params