-
Notifications
You must be signed in to change notification settings - Fork 2.5k
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
Showing
9 changed files
with
154 additions
and
155 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
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -1,37 +1,37 @@ | ||
| # model settings | ||
| embed_dims = [32, 64, 160, 256] | ||
| norm_cfg = dict(type='SyncBN', requires_grad=True) | ||
| model = dict( | ||
| type='EncoderDecoder', | ||
| pretrained=None, | ||
| backbone=dict( | ||
| type='SimplifiedMixTransformer', | ||
| in_channels=3, | ||
| embed_dims=embed_dims, | ||
| num_stages=4, | ||
| num_layers=[2, 2, 2, 2], | ||
| num_heads=[1, 2, 5, 8], | ||
| patch_sizes=[7, 3, 3, 3], | ||
| strides=[4, 2, 2, 2], | ||
| sr_ratios=[8, 4, 2, 1], | ||
| out_indices=(0, 1, 2, 3), | ||
| mlp_ratios=[8, 8, 4, 4], | ||
| qkv_bias=True, | ||
| drop_rate=0.0, | ||
| attn_drop_rate=0.0, | ||
| drop_path_rate=0.1, | ||
| norm_cfg=norm_cfg), | ||
| decode_head=dict( | ||
| type='DESTHead', | ||
| in_channels=[32, 64, 160, 256], | ||
| in_index=[0, 1, 2, 3], | ||
| channels=32, | ||
| dropout_ratio=0.1, | ||
| num_classes=19, | ||
| norm_cfg=norm_cfg, | ||
| align_corners=False, | ||
| loss_decode=dict( | ||
| type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)), | ||
| # model training and testing settings | ||
| train_cfg=dict(), | ||
| test_cfg=dict(mode='slide', crop_size=(1024, 1024), stride=(768, 768))) | ||
| # model settings | ||
| embed_dims = [32, 64, 160, 256] | ||
| norm_cfg = dict(type='SyncBN', requires_grad=True) | ||
| model = dict( | ||
| type='EncoderDecoder', | ||
| pretrained=None, | ||
| backbone=dict( | ||
| type='SimplifiedMixTransformer', | ||
| in_channels=3, | ||
| embed_dims=embed_dims, | ||
| num_stages=4, | ||
| num_layers=[2, 2, 2, 2], | ||
| num_heads=[1, 2, 5, 8], | ||
| patch_sizes=[7, 3, 3, 3], | ||
| strides=[4, 2, 2, 2], | ||
| sr_ratios=[8, 4, 2, 1], | ||
| out_indices=(0, 1, 2, 3), | ||
| mlp_ratios=[8, 8, 4, 4], | ||
| qkv_bias=True, | ||
| drop_rate=0.0, | ||
| attn_drop_rate=0.0, | ||
| drop_path_rate=0.1, | ||
| norm_cfg=norm_cfg), | ||
| decode_head=dict( | ||
| type='DESTHead', | ||
| in_channels=[32, 64, 160, 256], | ||
| in_index=[0, 1, 2, 3], | ||
| channels=32, | ||
| dropout_ratio=0.1, | ||
| num_classes=19, | ||
| norm_cfg=norm_cfg, | ||
| align_corners=False, | ||
| loss_decode=dict( | ||
| type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)), | ||
| # model training and testing settings | ||
| train_cfg=dict(), | ||
| test_cfg=dict(mode='slide', crop_size=(1024, 1024), stride=(768, 768))) |
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
60 changes: 30 additions & 30 deletions
60
configs/dest/dest_simpatt-b0_1024x1024_160k_cityscapes.py
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 |
|---|---|---|
| @@ -1,30 +1,30 @@ | ||
| _base_ = [ | ||
| '../_base_/models/dest_simpatt-b0.py', | ||
| '../_base_/datasets/cityscapes_1024x1024.py', | ||
| '../_base_/default_runtime.py', '../_base_/schedules/schedule_160k.py' | ||
| ] | ||
|
|
||
| optimizer = dict( | ||
| _delete_=True, | ||
| type='AdamW', | ||
| lr=0.00006, | ||
| betas=(0.9, 0.999), | ||
| weight_decay=0.01, | ||
| paramwise_cfg=dict( | ||
| custom_keys={ | ||
| 'pos_block': dict(decay_mult=0.), | ||
| 'norm': dict(decay_mult=0.), | ||
| 'head': dict(lr_mult=10.) | ||
| })) | ||
|
|
||
| lr_config = dict( | ||
| _delete_=True, | ||
| policy='poly', | ||
| warmup='linear', | ||
| warmup_iters=1500, | ||
| warmup_ratio=1e-6, | ||
| power=1.0, | ||
| min_lr=0.0, | ||
| by_epoch=False) | ||
|
|
||
| data = dict(samples_per_gpu=1, workers_per_gpu=1) | ||
| _base_ = [ | ||
| '../_base_/models/dest_simpatt-b0.py', | ||
| '../_base_/datasets/cityscapes_1024x1024.py', | ||
| '../_base_/default_runtime.py', '../_base_/schedules/schedule_160k.py' | ||
| ] | ||
|
|
||
| optimizer = dict( | ||
| _delete_=True, | ||
| type='AdamW', | ||
| lr=0.00006, | ||
| betas=(0.9, 0.999), | ||
| weight_decay=0.01, | ||
| paramwise_cfg=dict( | ||
| custom_keys={ | ||
| 'pos_block': dict(decay_mult=0.), | ||
| 'norm': dict(decay_mult=0.), | ||
| 'head': dict(lr_mult=10.) | ||
| })) | ||
|
|
||
| lr_config = dict( | ||
| _delete_=True, | ||
| policy='poly', | ||
| warmup='linear', | ||
| warmup_iters=1500, | ||
| warmup_ratio=1e-6, | ||
| power=1.0, | ||
| min_lr=0.0, | ||
| by_epoch=False) | ||
|
|
||
| data = dict(samples_per_gpu=1, workers_per_gpu=1) |
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 |
|---|---|---|
| @@ -1,9 +1,9 @@ | ||
| _base_ = ['./dest_simpatt-b0_1024x1024_160k_cityscapes.py'] | ||
|
|
||
| embed_dims = [64, 128, 250, 320] | ||
|
|
||
| model = dict( | ||
| type='EncoderDecoder', | ||
| pretrained=None, | ||
| backbone=dict(embed_dims=embed_dims), | ||
| decode_head=dict(in_channels=embed_dims, channels=64)) | ||
| _base_ = ['./dest_simpatt-b0_1024x1024_160k_cityscapes.py'] | ||
|
|
||
| embed_dims = [64, 128, 250, 320] | ||
|
|
||
| model = dict( | ||
| type='EncoderDecoder', | ||
| pretrained=None, | ||
| backbone=dict(embed_dims=embed_dims), | ||
| decode_head=dict(in_channels=embed_dims, channels=64)) |
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 |
|---|---|---|
| @@ -1,9 +1,9 @@ | ||
| _base_ = ['./dest_simpatt-b0_1024x1024_160k_cityscapes.py'] | ||
|
|
||
| embed_dims = [64, 128, 250, 320] | ||
|
|
||
| model = dict( | ||
| type='EncoderDecoder', | ||
| pretrained=None, | ||
| backbone=dict(embed_dims=embed_dims, num_layers=[3, 3, 6, 3]), | ||
| decode_head=dict(in_channels=embed_dims, channels=64)) | ||
| _base_ = ['./dest_simpatt-b0_1024x1024_160k_cityscapes.py'] | ||
|
|
||
| embed_dims = [64, 128, 250, 320] | ||
|
|
||
| model = dict( | ||
| type='EncoderDecoder', | ||
| pretrained=None, | ||
| backbone=dict(embed_dims=embed_dims, num_layers=[3, 3, 6, 3]), | ||
| decode_head=dict(in_channels=embed_dims, channels=64)) |
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 |
|---|---|---|
| @@ -1,52 +1,54 @@ | ||
| import torch | ||
| import torch.nn as nn | ||
|
|
||
| from mmcv.cnn import ConvModule | ||
| from mmseg.models.builder import HEADS | ||
| from mmseg.models.decode_heads.decode_head import BaseDecodeHead | ||
|
|
||
|
|
||
| @HEADS.register_module() | ||
| class DESTHead(BaseDecodeHead): | ||
|
|
||
| def __init__(self, interpolate_mode='bilinear', **kwargs): | ||
| super().__init__(input_transform='multiple_select', **kwargs) | ||
| self.interpolate_mode = interpolate_mode | ||
| num_inputs = len(self.in_channels) | ||
| assert num_inputs == len(self.in_index) | ||
| self.fuse_in_channels = self.in_channels.copy() | ||
| for i in range(num_inputs - 1): | ||
| self.fuse_in_channels[i] += self.fuse_in_channels[i + 1] | ||
| self.convs = nn.ModuleList() | ||
| for i in range(num_inputs): | ||
| self.convs.append( | ||
| ConvModule( | ||
| in_channels=self.in_channels[i], | ||
| out_channels=self.in_channels[i], | ||
| kernel_size=1, | ||
| stride=1, | ||
| act_cfg=self.act_cfg)) | ||
|
|
||
| self.fuse_convs = nn.ModuleList() | ||
| for i in range(num_inputs): | ||
| self.fuse_convs.append( | ||
| ConvModule( | ||
| in_channels=self.fuse_in_channels[i], | ||
| out_channels=self.in_channels[i], | ||
| kernel_size=3, | ||
| stride=1, | ||
| padding=1, | ||
| act_cfg=self.act_cfg)) | ||
|
|
||
| self.upsample = nn.ModuleList([ | ||
| nn.Sequential(nn.Upsample(scale_factor=2, mode=interpolate_mode)) | ||
| ] * len(self.in_channels)) | ||
|
|
||
| def forward(self, inputs): | ||
| for idx in reversed(range(len(inputs))): | ||
| x = self.convs[idx](inputs[idx]) | ||
| if idx != len(inputs) - 1: | ||
| x = torch.concat([feat, x], dim=1) | ||
| x = self.upsample[idx](x) | ||
| feat = self.fuse_convs[idx](x) | ||
| return self.cls_seg(feat) | ||
| # Copyright (c) OpenMMLab. All rights reserved. | ||
| import torch | ||
| import torch.nn as nn | ||
| from mmcv.cnn import ConvModule | ||
|
|
||
| from mmseg.models.builder import HEADS | ||
| from mmseg.models.decode_heads.decode_head import BaseDecodeHead | ||
|
|
||
|
|
||
| @HEADS.register_module() | ||
| class DESTHead(BaseDecodeHead): | ||
|
|
||
| def __init__(self, interpolate_mode='bilinear', **kwargs): | ||
| super().__init__(input_transform='multiple_select', **kwargs) | ||
| self.interpolate_mode = interpolate_mode | ||
| num_inputs = len(self.in_channels) | ||
| assert num_inputs == len(self.in_index) | ||
| self.fuse_in_channels = self.in_channels.copy() | ||
| for i in range(num_inputs - 1): | ||
| self.fuse_in_channels[i] += self.fuse_in_channels[i + 1] | ||
| self.convs = nn.ModuleList() | ||
| for i in range(num_inputs): | ||
| self.convs.append( | ||
| ConvModule( | ||
| in_channels=self.in_channels[i], | ||
| out_channels=self.in_channels[i], | ||
| kernel_size=1, | ||
| stride=1, | ||
| act_cfg=self.act_cfg)) | ||
|
|
||
| self.fuse_convs = nn.ModuleList() | ||
| for i in range(num_inputs): | ||
| self.fuse_convs.append( | ||
| ConvModule( | ||
| in_channels=self.fuse_in_channels[i], | ||
| out_channels=self.in_channels[i], | ||
| kernel_size=3, | ||
| stride=1, | ||
| padding=1, | ||
| act_cfg=self.act_cfg)) | ||
|
|
||
| self.upsample = nn.ModuleList([ | ||
| nn.Sequential(nn.Upsample(scale_factor=2, mode=interpolate_mode)) | ||
| ] * len(self.in_channels)) | ||
|
|
||
| def forward(self, inputs): | ||
| feat = None | ||
| for idx in reversed(range(len(inputs))): | ||
| x = self.convs[idx](inputs[idx]) | ||
| if idx != len(inputs) - 1: | ||
| x = torch.concat([feat, x], dim=1) | ||
| x = self.upsample[idx](x) | ||
| feat = self.fuse_convs[idx](x) | ||
| return self.cls_seg(feat) |
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