MiT_siamese.py

#############################################################
# NVIDIA Source Code License for SegFormer

# 1. Definitions

# “Licensor” means any person or entity that distributes its Work.

# “Software” means the original work of authorship made available under this License.

# “Work” means the Software and any additions to or derivative works of the Software that are made available under
# this License.

# The terms “reproduce,” “reproduction,” “derivative works,” and “distribution” have the meaning as provided under
# U.S. copyright law; provided, however, that for the purposes of this License, derivative works shall not include
# works that remain separable from, or merely link (or bind by name) to the interfaces of, the Work.

# Works, including the Software, are “made available” under this License by including in or with the Work either
# (a) a copyright notice referencing the applicability of this License to the Work, or (b) a copy of this License.

# 2. License Grant

# 2.1 Copyright Grant. Subject to the terms and conditions of this License, each Licensor grants to you a perpetual,
# worldwide, non-exclusive, royalty-free, copyright license to reproduce, prepare derivative works of, publicly
# display, publicly perform, sublicense and distribute its Work and any resulting derivative works in any form.

# 3. Limitations

# 3.1 Redistribution. You may reproduce or distribute the Work only if (a) you do so under this License, (b) you
# include a complete copy of this License with your distribution, and (c) you retain without modification any
# copyright, patent, trademark, or attribution notices that are present in the Work.

# 3.2 Derivative Works. You may specify that additional or different terms apply to the use, reproduction, and
# distribution of your derivative works of the Work (“Your Terms”) only if (a) Your Terms provide that the use
# limitation in Section 3.3 applies to your derivative works, and (b) you identify the specific derivative works
# that are subject to Your Terms. Notwithstanding Your Terms, this License (including the redistribution
# requirements in Section 3.1) will continue to apply to the Work itself.

# 3.3 Use Limitation. The Work and any derivative works thereof only may be used or intended for use
# non-commercially. Notwithstanding the foregoing, NVIDIA and its affiliates may use the Work and any derivative
# works commercially. As used herein, “non-commercially” means for research or evaluation purposes only.

# 3.4 Patent Claims. If you bring or threaten to bring a patent claim against any Licensor (including any claim,
# cross-claim or counterclaim in a lawsuit) to enforce any patents that you allege are infringed by any Work, then
# your rights under this License from such Licensor (including the grant in Section 2.1) will terminate immediately.

# 3.5 Trademarks. This License does not grant any rights to use any Licensor’s or its affiliates’ names, logos,
# or trademarks, except as necessary to reproduce the notices described in this License.

# 3.6 Termination. If you violate any term of this License, then your rights under this License (including the
# grant in Section 2.1) will terminate immediately.

# 4. Disclaimer of Warranty.

# THE WORK IS PROVIDED “AS IS” WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, EITHER EXPRESS OR IMPLIED, INCLUDING
# WARRANTIES OR CONDITIONS OF M ERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, TITLE OR NON-INFRINGEMENT. YOU
# BEAR THE RISK OF UNDERTAKING ANY ACTIVITIES UNDER THIS LICENSE.

# 5. Limitation of Liability.

# EXCEPT AS PROHIBITED BY APPLICABLE LAW, IN NO EVENT AND UNDER NO LEGAL THEORY, WHETHER IN TORT (INCLUDING
# NEGLIGENCE), CONTRACT, OR OTHERWISE SHALL ANY LICENSOR BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY DIRECT,
# INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES ARISING OUT OF OR RELATED TO THIS LICENSE, THE USE OR
# INABILITY TO USE THE WORK (INCLUDING BUT NOT LIMITED TO LOSS OF GOODWILL, BUSINESS INTERRUPTION, LOST PROFITS OR
# DATA, COMPUTER FAILURE OR MALFUNCTION, OR ANY OTHER COMM ERCIAL DAMAGES OR LOSSES), EVEN IF THE LICENSOR HAS BEEN
# ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.
#############################################################


import math
from functools import partial
from collections import OrderedDict
from abc import ABCMeta, abstractmethod

from mmcv.cnn import ConvModule
from timm.models.layers import DropPath, to_2tuple, trunc_normal_
import torch
import torch.nn as nn
import torch.nn.functional as F


def resize(input,
           size=None,
           scale_factor=None,
           mode='nearest',
           align_corners=None,
           warning=True):
    if warning:
        if size is not None and align_corners:
            input_h, input_w = tuple(int(x) for x in input.shape[2:])
            output_h, output_w = tuple(int(x) for x in size)
            if output_h > input_h or output_w > output_h:
                if ((output_h > 1 and output_w > 1 and input_h > 1
                     and input_w > 1) and (output_h - 1) % (input_h - 1)
                        and (output_w - 1) % (input_w - 1)):
                    warnings.warn(
                        f'When align_corners={align_corners}, '
                        'the output would more aligned if '
                        f'input size {(input_h, input_w)} is `x+1` and '
                        f'out size {(output_h, output_w)} is `nx+1`')
    return F.interpolate(input, size, scale_factor, mode, align_corners)


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.dwconv = DWConv(hidden_features)
        self.act = act_layer()
        self.fc2 = nn.Linear(hidden_features, out_features)
        self.drop = nn.Dropout(drop)

        self.apply(self._init_weights)

    def _init_weights(self, m):
        if isinstance(m, nn.Linear):
            trunc_normal_(m.weight, std=.02)
            if isinstance(m, nn.Linear) and m.bias is not None:
                nn.init.constant_(m.bias, 0)
        elif isinstance(m, nn.LayerNorm):
            nn.init.constant_(m.bias, 0)
            nn.init.constant_(m.weight, 1.0)
        elif isinstance(m, nn.Conv2d):
            fan_out = m.kernel_size[0] * m.kernel_size[1] * m.out_channels
            fan_out //= m.groups
            m.weight.data.normal_(0, math.sqrt(2.0 / fan_out))
            if m.bias is not None:
                m.bias.data.zero_()

    def forward(self, x, H, W):
        x = self.fc1(x)
        x = self.dwconv(x, H, W)
        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):
        super().__init__()
        assert dim % num_heads == 0, f"dim {dim} should be divided by num_heads {num_heads}."

        self.dim = dim
        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, stride=sr_ratio)
            self.norm = nn.LayerNorm(dim)

        self.apply(self._init_weights)

    def _init_weights(self, m):
        if isinstance(m, nn.Linear):
            trunc_normal_(m.weight, std=.02)
            if isinstance(m, nn.Linear) and m.bias is not None:
                nn.init.constant_(m.bias, 0)
        elif isinstance(m, nn.LayerNorm):
            nn.init.constant_(m.bias, 0)
            nn.init.constant_(m.weight, 1.0)
        elif isinstance(m, nn.Conv2d):
            fan_out = m.kernel_size[0] * m.kernel_size[1] * m.out_channels
            fan_out //= m.groups
            m.weight.data.normal_(0, math.sqrt(2.0 / fan_out))
            if m.bias is not None:
                m.bias.data.zero_()

    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.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, -1, 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
        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):
        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)
        # 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)

        self.apply(self._init_weights)

    def _init_weights(self, m):
        if isinstance(m, nn.Linear):
            trunc_normal_(m.weight, std=.02)
            if isinstance(m, nn.Linear) and m.bias is not None:
                nn.init.constant_(m.bias, 0)
        elif isinstance(m, nn.LayerNorm):
            nn.init.constant_(m.bias, 0)
            nn.init.constant_(m.weight, 1.0)
        elif isinstance(m, nn.Conv2d):
            fan_out = m.kernel_size[0] * m.kernel_size[1] * m.out_channels
            fan_out //= m.groups
            m.weight.data.normal_(0, math.sqrt(2.0 / fan_out))
            if m.bias is not None:
                m.bias.data.zero_()

    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), H, W))

        return x


class OverlapPatchEmbed(nn.Module):
    """ Image to Patch Embedding
    """

    def __init__(self, img_size=224, patch_size=7, stride=4, in_chans=3, embed_dim=768):
        super().__init__()
        img_size = to_2tuple(img_size)
        patch_size = to_2tuple(patch_size)

        self.img_size = img_size
        self.patch_size = patch_size
        self.H, self.W = img_size[0] // patch_size[0], img_size[1] // patch_size[1]
        self.num_patches = self.H * self.W
        self.proj = nn.Conv2d(in_chans, embed_dim, kernel_size=patch_size, stride=stride,
                              padding=(patch_size[0] // 2, patch_size[1] // 2))
        self.norm = nn.LayerNorm(embed_dim)

        self.apply(self._init_weights)

    def _init_weights(self, m):
        if isinstance(m, nn.Linear):
            trunc_normal_(m.weight, std=.02)
            if isinstance(m, nn.Linear) and m.bias is not None:
                nn.init.constant_(m.bias, 0)
        elif isinstance(m, nn.LayerNorm):
            nn.init.constant_(m.bias, 0)
            nn.init.constant_(m.weight, 1.0)
        elif isinstance(m, nn.Conv2d):
            fan_out = m.kernel_size[0] * m.kernel_size[1] * m.out_channels
            fan_out //= m.groups
            m.weight.data.normal_(0, math.sqrt(2.0 / fan_out))
            if m.bias is not None:
                m.bias.data.zero_()

    def forward(self, x):
        x = self.proj(x)
        _, _, H, W = x.shape
        x = x.flatten(2).transpose(1, 2)
        x = self.norm(x)

        return x, H, W


class MixVisionTransformer(nn.Module):
    def __init__(self, img_size=224, patch_size=16, 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., norm_layer=nn.LayerNorm,
                 depths=[3, 4, 6, 3], sr_ratios=[8, 4, 2, 1]):
        super().__init__()
        self.num_classes = num_classes
        self.depths = depths

        # patch_embed
        self.patch_embed1 = OverlapPatchEmbed(img_size=img_size, patch_size=7, stride=4, in_chans=in_chans,
                                              embed_dim=embed_dims[0])
        self.patch_embed1_refine = OverlapPatchEmbed(img_size=img_size, patch_size=7, stride=4, in_chans=4,
                                              embed_dim=embed_dims[0])
        self.patch_embed2 = OverlapPatchEmbed(img_size=img_size // 4, patch_size=3, stride=2, in_chans=embed_dims[0],
                                              embed_dim=embed_dims[1])
        self.patch_embed3 = OverlapPatchEmbed(img_size=img_size // 8, patch_size=3, stride=2, in_chans=embed_dims[1],
                                              embed_dim=embed_dims[2])
        self.patch_embed4 = OverlapPatchEmbed(img_size=img_size // 16, patch_size=3, stride=2, in_chans=embed_dims[2],
                                              embed_dim=embed_dims[3])

        # transformer encoder
        dpr = [x.item() for x in torch.linspace(0, drop_path_rate, sum(depths))]  # stochastic depth decay rule
        cur = 0
        self.block1 = nn.ModuleList([Block(
            dim=embed_dims[0], num_heads=num_heads[0], mlp_ratio=mlp_ratios[0], qkv_bias=qkv_bias, qk_scale=qk_scale,
            drop=drop_rate, attn_drop=attn_drop_rate, drop_path=dpr[cur + i], norm_layer=norm_layer,
            sr_ratio=sr_ratios[0])
            for i in range(depths[0])])
        
        self.block1_refine = nn.ModuleList([Block(
            dim=embed_dims[0], num_heads=num_heads[0], mlp_ratio=mlp_ratios[0], qkv_bias=qkv_bias, qk_scale=qk_scale,
            drop=drop_rate, attn_drop=attn_drop_rate, drop_path=dpr[cur + i], norm_layer=norm_layer,
            sr_ratio=sr_ratios[0])
            for i in range(depths[0])])
        self.norm1 = norm_layer(embed_dims[0])

        cur += depths[0]
        self.block2 = nn.ModuleList([Block(
            dim=embed_dims[1], num_heads=num_heads[1], mlp_ratio=mlp_ratios[1], qkv_bias=qkv_bias, qk_scale=qk_scale,
            drop=drop_rate, attn_drop=attn_drop_rate, drop_path=dpr[cur + i], norm_layer=norm_layer,
            sr_ratio=sr_ratios[1])
            for i in range(depths[1])])
        self.norm2 = norm_layer(embed_dims[1])

        cur += depths[1]
        self.block3 = nn.ModuleList([Block(
            dim=embed_dims[2], num_heads=num_heads[2], mlp_ratio=mlp_ratios[2], qkv_bias=qkv_bias, qk_scale=qk_scale,
            drop=drop_rate, attn_drop=attn_drop_rate, drop_path=dpr[cur + i], norm_layer=norm_layer,
            sr_ratio=sr_ratios[2])
            for i in range(depths[2])])
        self.norm3 = norm_layer(embed_dims[2])

        cur += depths[2]
        self.block4 = nn.ModuleList([Block(
            dim=embed_dims[3], num_heads=num_heads[3], mlp_ratio=mlp_ratios[3], qkv_bias=qkv_bias, qk_scale=qk_scale,
            drop=drop_rate, attn_drop=attn_drop_rate, drop_path=dpr[cur + i], norm_layer=norm_layer,
            sr_ratio=sr_ratios[3])
            for i in range(depths[3])])
        self.norm4 = norm_layer(embed_dims[3])

        # classification head
        # self.head = nn.Linear(embed_dims[3], num_classes) if num_classes > 0 else nn.Identity()

        self.apply(self._init_weights)

    def _init_weights(self, m):
        if isinstance(m, nn.Linear):
            trunc_normal_(m.weight, std=.02)
            if isinstance(m, nn.Linear) and m.bias is not None:
                nn.init.constant_(m.bias, 0)
        elif isinstance(m, nn.LayerNorm):
            nn.init.constant_(m.bias, 0)
            nn.init.constant_(m.weight, 1.0)
        elif isinstance(m, nn.Conv2d):
            fan_out = m.kernel_size[0] * m.kernel_size[1] * m.out_channels
            fan_out //= m.groups
            m.weight.data.normal_(0, math.sqrt(2.0 / fan_out))
            if m.bias is not None:
                m.bias.data.zero_()

    def init_weights(self, pretrained=None):
        if isinstance(pretrained, str):
            logger = get_root_logger()
            load_checkpoint(self, pretrained, map_location='cpu', strict=False, logger=logger)

    def reset_drop_path(self, drop_path_rate):
        dpr = [x.item() for x in torch.linspace(0, drop_path_rate, sum(self.depths))]
        cur = 0
        for i in range(self.depths[0]):
            self.block1[i].drop_path.drop_prob = dpr[cur + i]

        cur += self.depths[0]
        for i in range(self.depths[1]):
            self.block2[i].drop_path.drop_prob = dpr[cur + i]

        cur += self.depths[1]
        for i in range(self.depths[2]):
            self.block3[i].drop_path.drop_prob = dpr[cur + i]

        cur += self.depths[2]
        for i in range(self.depths[3]):
            self.block4[i].drop_path.drop_prob = dpr[cur + i]

    def freeze_patch_emb(self):
        self.patch_embed1.requires_grad = False

    @torch.jit.ignore
    def no_weight_decay(self):
        return {'pos_embed1', 'pos_embed2', 'pos_embed3', 'pos_embed4', 'cls_token'}  # has pos_embed may be better

    def get_classifier(self):
        return self.head

    def reset_classifier(self, num_classes, global_pool=''):
        self.num_classes = num_classes
        self.head = nn.Linear(self.embed_dim, num_classes) if num_classes > 0 else nn.Identity()

    def forward_features(self, x, x_r):
        B = x.shape[0]
        outs = []
        
        # stage 1
        x, H, W = self.patch_embed1(x)
        for i, blk in enumerate(self.block1):
            x = blk(x, H, W)
            
        x_r, H, W = self.patch_embed1_refine(x_r)
        for i, blk in enumerate(self.block1_refine):
            x_r = blk(x_r, H, W)
        
        x = self.norm1(x + x_r)
        x = x.reshape(B, H, W, -1).permute(0, 3, 1, 2).contiguous()
        outs.append(x)

        # stage 2
        x, H, W = self.patch_embed2(x)
        for i, blk in enumerate(self.block2):
            x = blk(x, H, W)
        x = self.norm2(x)
        x = x.reshape(B, H, W, -1).permute(0, 3, 1, 2).contiguous()
        outs.append(x)

        # stage 3
        x, H, W = self.patch_embed3(x)
        for i, blk in enumerate(self.block3):
            x = blk(x, H, W)
        x = self.norm3(x)
        x = x.reshape(B, H, W, -1).permute(0, 3, 1, 2).contiguous()
        outs.append(x)

        # stage 4
        x, H, W = self.patch_embed4(x)
        for i, blk in enumerate(self.block4):
            x = blk(x, H, W)
            
        x = self.norm4(x)
        x = x.reshape(B, H, W, -1).permute(0, 3, 1, 2).contiguous()
        outs.append(x)

        return outs

    def forward(self, x, x_r):
        x = self.forward_features(x, x_r)
        # x = self.head(x)

        return x


class DWConv(nn.Module):
    def __init__(self, dim=768):
        super(DWConv, self).__init__()
        self.dwconv = nn.Conv2d(dim, dim, 3, 1, 1, bias=True, groups=dim)

    def forward(self, x, H, W):
        B, N, C = x.shape
        x = x.transpose(1, 2).view(B, C, H, W)
        x = self.dwconv(x)
        x = x.flatten(2).transpose(1, 2)

        return x

        
class mit_b3(MixVisionTransformer):
    def __init__(self, **kwargs):
        super(mit_b3, self).__init__(
            patch_size=4, embed_dims=[64, 128, 320, 512], num_heads=[1, 2, 5, 8], mlp_ratios=[4, 4, 4, 4],
            qkv_bias=True, norm_layer=partial(nn.LayerNorm, eps=1e-6), depths=[3, 4, 18, 3], sr_ratios=[8, 4, 2, 1],
            drop_rate=0.0, drop_path_rate=0.1)

#######################################

class MLP(nn.Module):
    """
    Linear Embedding
    """
    def __init__(self, input_dim=2048, embed_dim=768):
        super().__init__()
        self.proj = nn.Linear(input_dim, embed_dim)

    def forward(self, x):
        x = x.flatten(2).transpose(1, 2)
        x = self.proj(x)
        return x


class BaseDecodeHead(nn.Module, metaclass=ABCMeta):
    """Base class for BaseDecodeHead.
    Args:
        in_channels (int|Sequence[int]): Input channels.
        channels (int): Channels after modules, before conv_seg.
        num_classes (int): Number of classes.
        dropout_ratio (float): Ratio of dropout layer. Default: 0.1.
        conv_cfg (dict|None): Config of conv layers. Default: None.
        norm_cfg (dict|None): Config of norm layers. Default: None.
        act_cfg (dict): Config of activation layers.
            Default: dict(type='ReLU')
        in_index (int|Sequence[int]): Input feature index. Default: -1
        input_transform (str|None): Transformation type of input features.
            Options: 'resize_concat', 'multiple_select', None.
            'resize_concat': Multiple feature maps will be resize to the
                same size as first one and than concat together.
                Usually used in FCN head of HRNet.
            'multiple_select': Multiple feature maps will be bundle into
                a list and passed into decode head.
            None: Only one select feature map is allowed.
            Default: None.
        loss_decode (dict): Config of decode loss.
            Default: dict(type='CrossEntropyLoss').
        ignore_index (int | None): The label index to be ignored. When using
            masked BCE loss, ignore_index should be set to None. Default: 255
        sampler (dict|None): The config of segmentation map sampler.
            Default: None.
        align_corners (bool): align_corners argument of F.interpolate.
            Default: False.
    """

    def __init__(self,
                 in_channels,
                 channels,
                 *,
                 num_classes,
                 dropout_ratio=0.1,
                 conv_cfg=None,
                 norm_cfg=None,
                 act_cfg=dict(type='ReLU'),
                 in_index=-1,
                 input_transform=None,
                 loss_decode=dict(
                     type='CrossEntropyLoss',
                     use_sigmoid=False,
                     loss_weight=1.0),
                 decoder_params=None,
                 ignore_index=255,
                 sampler=None,
                 align_corners=False):
        super(BaseDecodeHead, self).__init__()
        self._init_inputs(in_channels, in_index, input_transform)
        self.channels = channels
        self.num_classes = num_classes
        self.dropout_ratio = dropout_ratio
        self.conv_cfg = conv_cfg
        self.norm_cfg = norm_cfg
        self.act_cfg = act_cfg
        self.in_index = in_index
        self.ignore_index = ignore_index
        self.align_corners = align_corners

        if sampler is not None:
            self.sampler = build_pixel_sampler(sampler, context=self)
        else:
            self.sampler = None

        self.conv_seg = nn.Conv2d(channels, num_classes, kernel_size=1)
        if dropout_ratio > 0:
            self.dropout = nn.Dropout2d(dropout_ratio)
        else:
            self.dropout = None
        self.fp16_enabled = False

    def extra_repr(self):
        """Extra repr."""
        s = f'input_transform={self.input_transform}, ' \
            f'ignore_index={self.ignore_index}, ' \
            f'align_corners={self.align_corners}'
        return s

    def _init_inputs(self, in_channels, in_index, input_transform):
        """Check and initialize input transforms.
        The in_channels, in_index and input_transform must match.
        Specifically, when input_transform is None, only single feature map
        will be selected. So in_channels and in_index must be of type int.
        When input_transform
        Args:
            in_channels (int|Sequence[int]): Input channels.
            in_index (int|Sequence[int]): Input feature index.
            input_transform (str|None): Transformation type of input features.
                Options: 'resize_concat', 'multiple_select', None.
                'resize_concat': Multiple feature maps will be resize to the
                    same size as first one and than concat together.
                    Usually used in FCN head of HRNet.
                'multiple_select': Multiple feature maps will be bundle into
                    a list and passed into decode head.
                None: Only one select feature map is allowed.
        """

        if input_transform is not None:
            assert input_transform in ['resize_concat', 'multiple_select']
        self.input_transform = input_transform
        self.in_index = in_index
        if input_transform is not None:
            assert isinstance(in_channels, (list, tuple))
            assert isinstance(in_index, (list, tuple))
            assert len(in_channels) == len(in_index)
            if input_transform == 'resize_concat':
                self.in_channels = sum(in_channels)
            else:
                self.in_channels = in_channels
        else:
            assert isinstance(in_channels, int)
            assert isinstance(in_index, int)
            self.in_channels = in_channels

    def init_weights(self):
        """Initialize weights of classification layer."""
        normal_init(self.conv_seg, mean=0, std=0.01)

    def _transform_inputs(self, inputs):
        """Transform inputs for decoder.
        Args:
            inputs (list[Tensor]): List of multi-level img features.
        Returns:
            Tensor: The transformed inputs
        """

        if self.input_transform == 'resize_concat':
            inputs = [inputs[i] for i in self.in_index]
            upsampled_inputs = [
                resize(
                    input=x,
                    size=inputs[0].shape[2:],
                    mode='bilinear',
                    align_corners=self.align_corners) for x in inputs
            ]
            inputs = torch.cat(upsampled_inputs, dim=1)
        elif self.input_transform == 'multiple_select':
            inputs = [inputs[i] for i in self.in_index]
        else:
            inputs = inputs[self.in_index]

        return inputs

    @abstractmethod
    def forward(self, inputs):
        """Placeholder of forward function."""
        pass   
    
    
class LowLevelEncoder(nn.Module):
    def __init__(self, feat_dim=64, in_channel=3):
        super(LowLevelEncoder, self).__init__()
        self.conv1 = nn.Conv2d(3, feat_dim, kernel_size=7, stride=2, padding=3,
                               bias=False)
        self.bn1 = nn.BatchNorm2d(feat_dim)
        self.relu = nn.ReLU(inplace=True)
            
    def forward(self, x):
        x = self.conv1(x)
        x = self.bn1(x)
        x = self.relu(x)
        return x
        
        
################################################

class ResidualConvUnit(nn.Module):
    """Residual convolution module."""

    def __init__(self, features):
        """Init.
        Args:
            features (int): number of features
        """
        super().__init__()

        self.conv1 = nn.Conv2d(
            features, features, kernel_size=3, stride=1, padding=1, bias=True
        )

        self.conv2 = nn.Conv2d(
            features, features, kernel_size=3, stride=1, padding=1, bias=True
        )

        self.relu = torch.nn.ReLU(inplace=True)

    def forward(self, x):
        """Forward pass.
        Args:
            x (tensor): input
        Returns:
            tensor: output
        """
        out = self.relu(x)
        out = self.conv1(out)
        out = self.relu(out)
        out = self.conv2(out)

        return out + x


class FeatureFusionBlock(nn.Module):
    """Feature fusion block."""

    def __init__(self, features):
        """Init.
        Args:
            features (int): number of features
        """
        super(FeatureFusionBlock, self).__init__()

        self.resConfUnit1 = ResidualConvUnit(features)
        self.resConfUnit2 = ResidualConvUnit(features)

    def forward(self, *xs):
        """Forward pass.
        Returns:
            tensor: output
        """
        output = xs[0]

        if len(xs) == 2:
            output += self.resConfUnit1(xs[1])

        output = self.resConfUnit2(output)

        output = F.interpolate(
            output, scale_factor=2, mode="bilinear", align_corners=False
        )

        return output
    
class SegFormerHead(BaseDecodeHead):
    """
    SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers
    """
    def __init__(self, feature_strides, **kwargs):
        super(SegFormerHead, self).__init__(input_transform='multiple_select', **kwargs)
        assert len(feature_strides) == len(self.in_channels)
        assert min(feature_strides) == feature_strides[0]
        self.feature_strides = feature_strides

        c1_in_channels, c2_in_channels, c3_in_channels, c4_in_channels = self.in_channels

        decoder_params = kwargs['decoder_params']
        embedding_dim = decoder_params['embed_dim']

        self.linear_c4 = MLP(input_dim=c4_in_channels, embed_dim=embedding_dim)
        self.linear_c3 = MLP(input_dim=c3_in_channels, embed_dim=embedding_dim)
        self.linear_c2 = MLP(input_dim=c2_in_channels, embed_dim=embedding_dim)
        self.linear_c1 = MLP(input_dim=c1_in_channels, embed_dim=embedding_dim)
        
        
        self.linear_c4_proc =torch.nn.Conv2d(
            embedding_dim,
            256,
            kernel_size=3,
            stride=1,
            padding=1,
        )
        self.linear_c3_proc =torch.nn.Conv2d(
            embedding_dim,
            256,
            kernel_size=3,
            stride=1,
            padding=1,
        )
        self.linear_c2_proc =torch.nn.Conv2d(
            embedding_dim,
            256,
            kernel_size=3,
            stride=1,
            padding=1,
        )
        self.linear_c1_proc =torch.nn.Conv2d(
            embedding_dim,
            256,
            kernel_size=3,
            stride=1,
            padding=1,
        )
        
        self.fusion1 = FeatureFusionBlock(256)
        self.fusion2 = FeatureFusionBlock(256)
        self.fusion3 = FeatureFusionBlock(256)
        self.fusion4 = FeatureFusionBlock(256)
        
        
        self.conv_fuse_conv0 = ConvModule(
            in_channels=256 + 64,
            out_channels=64,
            kernel_size=3,
            padding=1,
        )
        
        self.conv_fuse_conv1 = ConvModule(
            in_channels=64,
            out_channels=32,
            kernel_size=3,
            padding=1,
        )

        self.linear_pred_depth_32 = nn.Conv2d(32, 1, kernel_size=1)

    def forward(self, inputs, input2):
        x = self._transform_inputs(inputs)  # len=4, 1/4,1/8,1/16,1/32
        c1, c2, c3, c4 = x

        ############## MLP decoder on C1-C4 ###########
        n, _, h, w = c4.shape

        _c4 = self.linear_c4(c4).permute(0,2,1).reshape(n, -1, c4.shape[2], c4.shape[3])
        _c4 = self.linear_c4_proc(_c4)
        _c4 = self.fusion4(_c4)

        _c3 = self.linear_c3(c3).permute(0,2,1).reshape(n, -1, c3.shape[2], c3.shape[3])
        _c3 = self.linear_c3_proc(_c3)
        _c3 = self.fusion3(_c4, _c3)

        _c2 = self.linear_c2(c2).permute(0,2,1).reshape(n, -1, c2.shape[2], c2.shape[3])
        _c2 = self.linear_c2_proc(_c2)
        _c2 = self.fusion2(_c3, _c2)

        _c1 = self.linear_c1(c1).permute(0,2,1).reshape(n, -1, c1.shape[2], c1.shape[3])
        _c1 = self.linear_c1_proc(_c1)
        _c1 = self.fusion1(_c2, _c1)

        x = torch.cat([_c1, input2], dim=1)
        x = self.conv_fuse_conv0(x)
        x = F.interpolate(x, scale_factor=2, mode='bilinear', align_corners=False)
        x = self.conv_fuse_conv1(x)
        
        x = self.linear_pred_depth_32(x)
        
        return x
    

################################################

class MiTDepthModel(nn.Module):
    def __init__(self, **kwargs):
        super(MiTDepthModel, self).__init__()
        self.backbone = mit_b3()
        self.decode_head = SegFormerHead(
            in_channels=[64, 128, 320, 512],
            in_index=[0, 1, 2, 3],
            feature_strides=[4, 8, 16, 32],
            channels=128,
            dropout_ratio=0.1,
            num_classes=150,
            norm_cfg=dict(type='SyncBN', requires_grad=True),
            align_corners=False,
            decoder_params=dict(embed_dim=768)
        )
        self.ll_enc = LowLevelEncoder()

    def forward(self, x, x_r):
        x1 = self.backbone(x, x_r)
        x2 = self.ll_enc(x)
        x = self.decode_head(x1,x2)
        return x

############################################################


class RelDepthModel(nn.Module):
    def __init__(self):
        super(RelDepthModel, self).__init__()
        self.depth_model = MiTDepthModel()
        self.depth_model.to(memory_format=torch.channels_last)

    
    def forward(self, data, is_train=True):
        ''' 
        Depth refinement with binary mask and RGB image.
        data['input_rgb']: HxWx3
        data['input_depth']: HxWx1 (masked depth)
        data['mask']: HxWx1 (binary mask)
        '''
        self.input_rgb = data['input_rgb'].cuda()
        self.input_rgb.to(memory_format=torch.channels_last)
        
        self.mask_fg = data['mask'].cuda()
        self.mask_fg.to(memory_format=torch.channels_last)
        self.mask_bg = 1 - self.mask_fg
        
        self.input_depth = data['input_depth'].cuda()
        self.input_depth.to(memory_format=torch.channels_last) 
        
        self.input_depth_fg_br = torch.cat((self.input_depth, self.mask_fg, self.mask_fg), dim=1)
        self.input_rgb_fg_br = torch.cat((self.input_rgb, self.mask_fg), dim=1)
        
        self.input_depth_bg_br = torch.cat((self.input_depth, self.mask_bg, self.mask_bg), dim=1)
        self.input_rgb_bg_br = torch.cat((self.input_rgb, self.mask_bg), dim=1)
        
        self.pred_fg = self.depth_model(self.input_depth_fg_br, self.input_rgb_fg_br)
        self.pred_bg = self.depth_model(self.input_depth_bg_br, self.input_rgb_bg_br)
        self.pred = self.pred_fg * self.mask_fg + self.pred_bg * self.mask_bg
        
        return self.pred, self.pred_fg, self.pred_bg