import logging import numpy as np import torch import torch.nn as nn from .backbone import Backbone from .utils import ( PatchEmbed, add_decomposed_rel_pos, get_abs_pos, window_partition, window_unpartition, ) logger = logging.getLogger(__name__) __all__ = ["MViT"] def attention_pool(x, pool, norm=None): # (B, H, W, C) -> (B, C, H, W) x = x.permute(0, 3, 1, 2) x = pool(x) # (B, C, H1, W1) -> (B, H1, W1, C) x = x.permute(0, 2, 3, 1) if norm: x = norm(x) return x class MultiScaleAttention(nn.Module): """Multiscale Multi-head Attention block.""" def __init__( self, dim, dim_out, num_heads, qkv_bias=True, norm_layer=nn.LayerNorm, pool_kernel=(3, 3), stride_q=1, stride_kv=1, residual_pooling=True, window_size=0, use_rel_pos=False, rel_pos_zero_init=True, input_size=None, ): """ Args: dim (int): Number of input channels. dim_out (int): Number of output channels. num_heads (int): Number of attention heads. qkv_bias (bool: If True, add a learnable bias to query, key, value. norm_layer (nn.Module): Normalization layer. pool_kernel (tuple): kernel size for qkv pooling layers. stride_q (int): stride size for q pooling layer. stride_kv (int): stride size for kv pooling layer. residual_pooling (bool): If true, enable residual pooling. use_rel_pos (bool): If True, add relative postional embeddings to the attention map. rel_pos_zero_init (bool): If True, zero initialize relative positional parameters. input_size (int or None): Input resolution. """ super().__init__() self.num_heads = num_heads head_dim = dim_out // num_heads self.scale = head_dim**-0.5 self.qkv = nn.Linear(dim, dim_out * 3, bias=qkv_bias) self.proj = nn.Linear(dim_out, dim_out) # qkv pooling pool_padding = [k // 2 for k in pool_kernel] dim_conv = dim_out // num_heads self.pool_q = nn.Conv2d( dim_conv, dim_conv, pool_kernel, stride=stride_q, padding=pool_padding, groups=dim_conv, bias=False, ) self.norm_q = norm_layer(dim_conv) self.pool_k = nn.Conv2d( dim_conv, dim_conv, pool_kernel, stride=stride_kv, padding=pool_padding, groups=dim_conv, bias=False, ) self.norm_k = norm_layer(dim_conv) self.pool_v = nn.Conv2d( dim_conv, dim_conv, pool_kernel, stride=stride_kv, padding=pool_padding, groups=dim_conv, bias=False, ) self.norm_v = norm_layer(dim_conv) self.window_size = window_size if window_size: self.q_win_size = window_size // stride_q self.kv_win_size = window_size // stride_kv self.residual_pooling = residual_pooling self.use_rel_pos = use_rel_pos if self.use_rel_pos: # initialize relative positional embeddings assert input_size[0] == input_size[1] size = input_size[0] rel_dim = 2 * max(size // stride_q, size // stride_kv) - 1 self.rel_pos_h = nn.Parameter(torch.zeros(rel_dim, head_dim)) self.rel_pos_w = nn.Parameter(torch.zeros(rel_dim, head_dim)) if not rel_pos_zero_init: nn.init.trunc_normal_(self.rel_pos_h, std=0.02) nn.init.trunc_normal_(self.rel_pos_w, std=0.02) def forward(self, x): B, H, W, _ = x.shape # qkv with shape (3, B, nHead, H, W, C) qkv = self.qkv(x).reshape(B, H, W, 3, self.num_heads, -1).permute(3, 0, 4, 1, 2, 5) # q, k, v with shape (B * nHead, H, W, C) q, k, v = qkv.reshape(3, B * self.num_heads, H, W, -1).unbind(0) q = attention_pool(q, self.pool_q, self.norm_q) k = attention_pool(k, self.pool_k, self.norm_k) v = attention_pool(v, self.pool_v, self.norm_v) ori_q = q if self.window_size: q, q_hw_pad = window_partition(q, self.q_win_size) k, kv_hw_pad = window_partition(k, self.kv_win_size) v, _ = window_partition(v, self.kv_win_size) q_hw = (self.q_win_size, self.q_win_size) kv_hw = (self.kv_win_size, self.kv_win_size) else: q_hw = q.shape[1:3] kv_hw = k.shape[1:3] q = q.view(q.shape[0], np.prod(q_hw), -1) k = k.view(k.shape[0], np.prod(kv_hw), -1) v = v.view(v.shape[0], np.prod(kv_hw), -1) attn = (q * self.scale) @ k.transpose(-2, -1) if self.use_rel_pos: attn = add_decomposed_rel_pos(attn, q, self.rel_pos_h, self.rel_pos_w, q_hw, kv_hw) attn = attn.softmax(dim=-1) x = attn @ v x = x.view(x.shape[0], q_hw[0], q_hw[1], -1) if self.window_size: x = window_unpartition(x, self.q_win_size, q_hw_pad, ori_q.shape[1:3]) if self.residual_pooling: x += ori_q H, W = x.shape[1], x.shape[2] x = x.view(B, self.num_heads, H, W, -1).permute(0, 2, 3, 1, 4).reshape(B, H, W, -1) x = self.proj(x) return x class MultiScaleBlock(nn.Module): """Multiscale Transformer blocks""" def __init__( self, dim, dim_out, num_heads, mlp_ratio=4.0, qkv_bias=True, drop_path=0.0, norm_layer=nn.LayerNorm, act_layer=nn.GELU, qkv_pool_kernel=(3, 3), stride_q=1, stride_kv=1, residual_pooling=True, window_size=0, use_rel_pos=False, rel_pos_zero_init=True, input_size=None, ): """ Args: dim (int): Number of input channels. dim_out (int): Number of output channels. num_heads (int): Number of attention heads in the MViT block. mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. qkv_bias (bool): If True, add a learnable bias to query, key, value. drop_path (float): Stochastic depth rate. norm_layer (nn.Module): Normalization layer. act_layer (nn.Module): Activation layer. qkv_pool_kernel (tuple): kernel size for qkv pooling layers. stride_q (int): stride size for q pooling layer. stride_kv (int): stride size for kv pooling layer. residual_pooling (bool): If true, enable residual pooling. window_size (int): Window size for window attention blocks. If it equals 0, then not use window attention. use_rel_pos (bool): If True, add relative postional embeddings to the attention map. rel_pos_zero_init (bool): If True, zero initialize relative positional parameters. input_size (int or None): Input resolution. """ super().__init__() self.norm1 = norm_layer(dim) self.attn = MultiScaleAttention( dim, dim_out, num_heads=num_heads, qkv_bias=qkv_bias, norm_layer=norm_layer, pool_kernel=qkv_pool_kernel, stride_q=stride_q, stride_kv=stride_kv, residual_pooling=residual_pooling, window_size=window_size, use_rel_pos=use_rel_pos, rel_pos_zero_init=rel_pos_zero_init, input_size=input_size, ) from timm.models.layers import DropPath, Mlp self.drop_path = DropPath(drop_path) if drop_path > 0.0 else nn.Identity() self.norm2 = norm_layer(dim_out) self.mlp = Mlp( in_features=dim_out, hidden_features=int(dim_out * mlp_ratio), out_features=dim_out, act_layer=act_layer, ) if dim != dim_out: self.proj = nn.Linear(dim, dim_out) if stride_q > 1: kernel_skip = stride_q + 1 padding_skip = int(kernel_skip // 2) self.pool_skip = nn.MaxPool2d(kernel_skip, stride_q, padding_skip, ceil_mode=False) def forward(self, x): x_norm = self.norm1(x) x_block = self.attn(x_norm) if hasattr(self, "proj"): x = self.proj(x_norm) if hasattr(self, "pool_skip"): x = attention_pool(x, self.pool_skip) x = x + self.drop_path(x_block) x = x + self.drop_path(self.mlp(self.norm2(x))) return x class MViT(Backbone): """ This module implements Multiscale Vision Transformer (MViT) backbone in :paper:'mvitv2'. """ def __init__( self, img_size=224, patch_kernel=(7, 7), patch_stride=(4, 4), patch_padding=(3, 3), in_chans=3, embed_dim=96, depth=16, num_heads=1, last_block_indexes=(0, 2, 11, 15), qkv_pool_kernel=(3, 3), adaptive_kv_stride=4, adaptive_window_size=56, residual_pooling=True, mlp_ratio=4.0, qkv_bias=True, drop_path_rate=0.0, norm_layer=nn.LayerNorm, act_layer=nn.GELU, use_abs_pos=False, use_rel_pos=True, rel_pos_zero_init=True, use_act_checkpoint=False, pretrain_img_size=224, pretrain_use_cls_token=True, out_features=("scale2", "scale3", "scale4", "scale5"), ): """ Args: img_size (int): Input image size. patch_kernel (tuple): kernel size for patch embedding. patch_stride (tuple): stride size for patch embedding. patch_padding (tuple): padding size for patch embedding. in_chans (int): Number of input image channels. embed_dim (int): Patch embedding dimension. depth (int): Depth of MViT. num_heads (int): Number of base attention heads in each MViT block. last_block_indexes (tuple): Block indexes for last blocks in each stage. qkv_pool_kernel (tuple): kernel size for qkv pooling layers. adaptive_kv_stride (int): adaptive stride size for kv pooling. adaptive_window_size (int): adaptive window size for window attention blocks. residual_pooling (bool): If true, enable residual pooling. mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. qkv_bias (bool): If True, add a learnable bias to query, key, value. drop_path_rate (float): Stochastic depth rate. norm_layer (nn.Module): Normalization layer. act_layer (nn.Module): Activation layer. use_abs_pos (bool): If True, use absolute positional embeddings. use_rel_pos (bool): If True, add relative postional embeddings to the attention map. rel_pos_zero_init (bool): If True, zero initialize relative positional parameters. window_size (int): Window size for window attention blocks. use_act_checkpoint (bool): If True, use activation checkpointing. pretrain_img_size (int): input image size for pretraining models. pretrain_use_cls_token (bool): If True, pretrainig models use class token. out_features (tuple): name of the feature maps from each stage. """ super().__init__() self.pretrain_use_cls_token = pretrain_use_cls_token self.patch_embed = PatchEmbed( kernel_size=patch_kernel, stride=patch_stride, padding=patch_padding, in_chans=in_chans, embed_dim=embed_dim, ) if use_abs_pos: # Initialize absoluate positional embedding with pretrain image size. num_patches = (pretrain_img_size // patch_stride[0]) * ( pretrain_img_size // patch_stride[1] ) num_positions = (num_patches + 1) if pretrain_use_cls_token else num_patches self.pos_embed = nn.Parameter(torch.zeros(1, num_positions, embed_dim)) else: self.pos_embed = None # stochastic depth decay rule dpr = [x.item() for x in torch.linspace(0, drop_path_rate, depth)] dim_out = embed_dim stride_kv = adaptive_kv_stride window_size = adaptive_window_size input_size = (img_size // patch_stride[0], img_size // patch_stride[1]) stage = 2 stride = patch_stride[0] self._out_feature_strides = {} self._out_feature_channels = {} self.blocks = nn.ModuleList() for i in range(depth): # Multiply stride_kv by 2 if it's the last block of stage2 and stage3. if i == last_block_indexes[1] or i == last_block_indexes[2]: stride_kv_ = stride_kv * 2 else: stride_kv_ = stride_kv # hybrid window attention: global attention in last three stages. window_size_ = 0 if i in last_block_indexes[1:] else window_size block = MultiScaleBlock( dim=embed_dim, dim_out=dim_out, num_heads=num_heads, mlp_ratio=mlp_ratio, qkv_bias=qkv_bias, drop_path=dpr[i], norm_layer=norm_layer, qkv_pool_kernel=qkv_pool_kernel, stride_q=2 if i - 1 in last_block_indexes else 1, stride_kv=stride_kv_, residual_pooling=residual_pooling, window_size=window_size_, use_rel_pos=use_rel_pos, rel_pos_zero_init=rel_pos_zero_init, input_size=input_size, ) if use_act_checkpoint: # TODO: use torch.utils.checkpoint from fairscale.nn.checkpoint import checkpoint_wrapper block = checkpoint_wrapper(block) self.blocks.append(block) embed_dim = dim_out if i in last_block_indexes: name = f"scale{stage}" if name in out_features: self._out_feature_channels[name] = dim_out self._out_feature_strides[name] = stride self.add_module(f"{name}_norm", norm_layer(dim_out)) dim_out *= 2 num_heads *= 2 stride_kv = max(stride_kv // 2, 1) stride *= 2 stage += 1 if i - 1 in last_block_indexes: window_size = window_size // 2 input_size = [s // 2 for s in input_size] self._out_features = out_features self._last_block_indexes = last_block_indexes if self.pos_embed is not None: nn.init.trunc_normal_(self.pos_embed, std=0.02) self.apply(self._init_weights) def _init_weights(self, m): if isinstance(m, nn.Linear): nn.init.trunc_normal_(m.weight, std=0.02) 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) def forward(self, x): x = self.patch_embed(x) if self.pos_embed is not None: x = x + get_abs_pos(self.pos_embed, self.pretrain_use_cls_token, x.shape[1:3]) outputs = {} stage = 2 for i, blk in enumerate(self.blocks): x = blk(x) if i in self._last_block_indexes: name = f"scale{stage}" if name in self._out_features: x_out = getattr(self, f"{name}_norm")(x) outputs[name] = x_out.permute(0, 3, 1, 2) stage += 1 return outputs