forked from sithu31296/semantic-segmentation
-
Notifications
You must be signed in to change notification settings - Fork 0
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
- Loading branch information
1 parent
8b4ffbe
commit e69675f
Showing
3 changed files
with
195 additions
and
3 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,180 @@ | ||
| import torch | ||
| from torch import nn, Tensor | ||
| from semseg.models.layers import DropPath | ||
|
|
||
|
|
||
| class MLP(nn.Module): | ||
| def __init__(self, dim, hidden_dim, out_dim=None) -> None: | ||
| super().__init__() | ||
| out_dim = out_dim or dim | ||
| self.fc1 = nn.Linear(dim, hidden_dim) | ||
| self.act = nn.GELU() | ||
| self.fc2 = nn.Linear(hidden_dim, out_dim) | ||
|
|
||
| def forward(self, x: Tensor) -> Tensor: | ||
| return self.fc2(self.act(self.fc1(x))) | ||
|
|
||
|
|
||
| class CMLP(nn.Module): | ||
| def __init__(self, dim, hidden_dim, out_dim=None) -> None: | ||
| super().__init__() | ||
| out_dim = out_dim or dim | ||
| self.fc1 = nn.Conv2d(dim, hidden_dim, 1) | ||
| self.act = nn.GELU() | ||
| self.fc2 = nn.Conv2d(hidden_dim, out_dim, 1) | ||
|
|
||
| def forward(self, x: Tensor) -> Tensor: | ||
| return self.fc2(self.act(self.fc1(x))) | ||
|
|
||
|
|
||
| class Attention(nn.Module): | ||
| def __init__(self, dim, num_heads=8) -> None: | ||
| super().__init__() | ||
| self.num_heads = num_heads | ||
| self.scale = (dim // num_heads) ** -0.5 | ||
| self.qkv = nn.Linear(dim, dim*3) | ||
| self.proj = nn.Linear(dim, dim) | ||
|
|
||
| def forward(self, x: Tensor) -> Tensor: | ||
| B, N, C = x.shape | ||
| q, k, v = self.qkv(x).reshape(B, N, 3, self.num_heads, C//self.num_heads).permute(2, 0, 3, 1, 4) | ||
| attn = (q @ k.transpose(-2, -1)) * self.scale | ||
| attn = attn.softmax(dim=-1) | ||
|
|
||
| x = (attn @ v).transpose(1, 2).reshape(B, N, C) | ||
| x = self.proj(x) | ||
| return x | ||
|
|
||
|
|
||
| class CBlock(nn.Module): | ||
| def __init__(self, dim, dpr=0.): | ||
| super().__init__() | ||
| self.pos_embed = nn.Conv2d(dim, dim, 3, 1, 1, groups=dim) | ||
| self.norm1 = nn.BatchNorm2d(dim) | ||
| self.conv1 = nn.Conv2d(dim, dim, 1) | ||
| self.conv2 = nn.Conv2d(dim, dim, 1) | ||
| self.attn = nn.Conv2d(dim, dim, 5, 1, 2, groups=dim) | ||
| self.drop_path = DropPath(dpr) if dpr > 0. else nn.Identity() | ||
| self.norm2 = nn.BatchNorm2d(dim) | ||
| self.mlp = CMLP(dim, int(dim*4)) | ||
|
|
||
| def forward(self, x: Tensor) -> Tensor: | ||
| x = x + self.pos_embed(x) | ||
| x = x + self.drop_path(self.conv2(self.attn(self.conv1(self.norm1(x))))) | ||
| x = x + self.drop_path(self.mlp(self.norm2(x))) | ||
| return x | ||
|
|
||
|
|
||
| class SABlock(nn.Module): | ||
| def __init__(self, dim, num_heads, dpr=0.) -> None: | ||
| super().__init__() | ||
| self.pos_embed = nn.Conv2d(dim, dim, 3, 1, 1, groups=dim) | ||
| self.norm1 = nn.LayerNorm(dim) | ||
| self.attn = Attention(dim, num_heads) | ||
| self.drop_path = DropPath(dpr) if dpr > 0. else nn.Identity() | ||
| self.norm2 = nn.LayerNorm(dim) | ||
| self.mlp = MLP(dim, int(dim*4)) | ||
|
|
||
| def forward(self, x: Tensor) -> Tensor: | ||
| x = x + self.pos_embed(x) | ||
| B, N, H, W = x.shape | ||
| x = x.flatten(2).transpose(1, 2) | ||
| x = x + self.drop_path(self.attn(self.norm1(x))) | ||
| x = x + self.drop_path(self.mlp(self.norm2(x))) | ||
| x = x.transpose(1, 2).reshape(B, N, H, W) | ||
| return x | ||
|
|
||
|
|
||
| class PatchEmbed(nn.Module): | ||
| def __init__(self, patch_size=16, in_ch=3, embed_dim=768) -> None: | ||
| super().__init__() | ||
| self.norm = nn.LayerNorm(embed_dim) | ||
| self.proj = nn.Conv2d(in_ch, embed_dim, patch_size, patch_size) | ||
|
|
||
| def forward(self, x: Tensor) -> Tensor: | ||
| x = self.proj(x) | ||
| B, C, H, W = x.shape | ||
| x = x.flatten(2).transpose(1, 2) | ||
| x = self.norm(x) | ||
| x = x.reshape(B, H, W, -1).permute(0, 3, 1, 2).contiguous() | ||
| return x | ||
|
|
||
|
|
||
| uniformer_settings = { | ||
| 'S': [3, 4, 8, 3], # [depth] | ||
| 'B': [5, 8, 20, 7] | ||
| } | ||
|
|
||
|
|
||
| class UniFormer(nn.Module): | ||
| def __init__(self, model_name: str = 'S') -> None: | ||
| super().__init__() | ||
| assert model_name in uniformer_settings.keys(), f"UniFormer model name should be in {list(uniformer_settings.keys())}" | ||
| depth = uniformer_settings[model_name] | ||
|
|
||
| head_dim = 64 | ||
| drop_path_rate = 0. | ||
| embed_dims = [64, 128, 320, 512] | ||
|
|
||
| for i in range(4): | ||
| self.add_module(f"patch_embed{i+1}", PatchEmbed(4 if i == 0 else 2, 3 if i == 0 else embed_dims[i-1], embed_dims[i])) | ||
| self.add_module(f"norm{i+1}", nn.LayerNorm(embed_dims[i])) | ||
|
|
||
| self.pos_drop = nn.Dropout(0.) | ||
| dpr = [x.item() for x in torch.linspace(0, drop_path_rate, sum(depth))] | ||
| num_heads = [dim // head_dim for dim in embed_dims] | ||
|
|
||
| self.blocks1 = nn.ModuleList([ | ||
| CBlock(embed_dims[0], dpr[i]) | ||
| for i in range(depth[0])]) | ||
|
|
||
| self.blocks2 = nn.ModuleList([ | ||
| CBlock(embed_dims[1], dpr[i+depth[0]]) | ||
| for i in range(depth[1])]) | ||
|
|
||
| self.blocks3 = nn.ModuleList([ | ||
| SABlock(embed_dims[2], num_heads[2], dpr[i+depth[0]+depth[1]]) | ||
| for i in range(depth[2])]) | ||
|
|
||
| self.blocks4 = nn.ModuleList([ | ||
| SABlock(embed_dims[3], num_heads[3], dpr[i+depth[0]+depth[1]+depth[2]]) | ||
| for i in range(depth[3])]) | ||
|
|
||
|
|
||
| def forward(self, x: torch.Tensor): | ||
| outs = [] | ||
|
|
||
| x = self.patch_embed1(x) | ||
| x = self.pos_drop(x) | ||
| for blk in self.blocks1: | ||
| x = blk(x) | ||
| x_out = self.norm1(x.permute(0, 2, 3, 1)) | ||
| outs.append(x_out.permute(0, 3, 1, 2)) | ||
|
|
||
| x = self.patch_embed2(x) | ||
| for blk in self.blocks2: | ||
| x = blk(x) | ||
| x_out = self.norm2(x.permute(0, 2, 3, 1)) | ||
| outs.append(x_out.permute(0, 3, 1, 2)) | ||
|
|
||
| x = self.patch_embed3(x) | ||
| for blk in self.blocks3: | ||
| x = blk(x) | ||
| x_out = self.norm3(x.permute(0, 2, 3, 1)) | ||
| outs.append(x_out.permute(0, 3, 1, 2)) | ||
|
|
||
| x = self.patch_embed4(x) | ||
| for blk in self.blocks4: | ||
| x = blk(x) | ||
| x_out = self.norm4(x.permute(0, 2, 3, 1)) | ||
| outs.append(x_out.permute(0, 3, 1, 2)) | ||
|
|
||
| return outs | ||
|
|
||
| if __name__ == '__main__': | ||
| model = UniFormer('S') | ||
| model.load_state_dict(torch.load('C:\\Users\\sithu\\Documents\\weights\\backbones\\uniformer\\uniformer_small_in1k.pth', map_location='cpu')['model'], strict=False) | ||
| x = torch.randn(1, 3, 224, 224) | ||
| feats = model(x) | ||
| for y in feats: | ||
| print(y.shape) |