Skip to content

Codex/generate cm dgseg multi modal segmentation code #3868

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
zhang317422 wants to merge 2 commits into
base: main
Choose a base branch
from
Open

Codex/generate cm dgseg multi modal segmentation code #3868

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
03e0467
Add CM-DGSeg dual-stream segmentation components
zhang317422 Nov 7, 2025
6522f09
2
zhang317422 Dec 2, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
197 changes: 197 additions & 0 deletions configs/cm-dgseg/cm_dgseg_b0_cityscapes.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,197 @@
# Copyright (c) OpenMMLab. All rights reserved.
"""CM-DGSeg B0 configuration for Cityscapes.

Run command:
python tools/train.py configs/cm-dgseg/cm_dgseg_b0_cityscapes.py
python tools/test.py configs/cm-dgseg/cm_dgseg_b0_cityscapes.py <checkpoint> --eval mIoU
"""

_base_ = [
'../_base_/datasets/cityscapes.py',
'../_base_/default_runtime.py',
]

crop_size = (512, 1024)
num_classes = 19
norm_cfg = dict(type='BN', requires_grad=True)
data_preprocessor = dict(
type='SegDataPreProcessor',
mean=[123.675, 116.28, 103.53, 0.0],
std=[58.395, 57.12, 57.375, 1.0],
bgr_to_rgb=True,
pad_val=0,
seg_pad_val=255,
size=crop_size)

model = dict(
type='DualStreamEncoderDecoder',
data_preprocessor=data_preprocessor,
backbone_rgb=dict(
type='MixVisionTransformer',
init_cfg=dict(type='Pretrained', checkpoint='https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segformer/mit_b0_20220624-7e0fe6dd.pth'),
in_channels=3,
embed_dims=32,
num_stages=4,
num_layers=[2, 2, 2, 2],
num_heads=[1, 2, 5, 8],
patch_sizes=[7, 3, 3, 3],
sr_ratios=[8, 4, 2, 1],
out_indices=(0, 1, 2, 3),
mlp_ratio=4,
qkv_bias=True,
drop_rate=0.0,
attn_drop_rate=0.0,
drop_path_rate=0.1),
backbone_disp=dict(
type='MixVisionTransformer',
in_channels=1,
embed_dims=16,
num_stages=4,
num_layers=[2, 2, 2, 2],
num_heads=[1, 1, 3, 4],
patch_sizes=[7, 3, 3, 3],
sr_ratios=[8, 4, 2, 1],
out_indices=(0, 1, 2, 3),
mlp_ratio=4,
qkv_bias=True,
drop_rate=0.0,
attn_drop_rate=0.0,
drop_path_rate=0.05),
decode_head=dict(
type='CMFSegFormerHead',
in_channels=[32, 64, 160, 256],
disp_in_channels=[16, 16, 48, 64],
in_index=[0, 1, 2, 3],
channels=256,
dropout_ratio=0.1,
num_classes=num_classes,
norm_cfg=norm_cfg,
align_corners=False,
with_boundary=False,
ignore_index=255,
loss_decode=[
dict(
type='CrossEntropyLoss',
use_sigmoid=False,
loss_weight=1.0,
avg_non_ignore=True
),
dict(
type='LovaszLoss',
per_image=False,
reduction='none',
loss_weight=0.4
)
]
),
train_cfg=dict(),
test_cfg=dict(mode='slide', crop_size=crop_size, stride=(768, 768)))

# Data pipelines: disable photometric distortion to keep disparity channel clean.
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='LoadCityscapesDisparity'),
dict(type='ConcatRGBDispTo4Ch', keep_rgb_dtype=False),
dict(type='LoadAnnotations'),
dict(type='RandomResize', scale=(2048, 1024), ratio_range=(0.5, 2.0), keep_ratio=True),
dict(type='RandomCrop', crop_size=crop_size, cat_max_ratio=0.75),
dict(type='RandomFlip', prob=0.5),
dict(type='PackSegInputs')
]

test_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='LoadCityscapesDisparity'),
dict(type='ConcatRGBDispTo4Ch', keep_rgb_dtype=False, delete_disp=True),
dict(type='Resize', scale=(2048, 1024), keep_ratio=True),
dict(type='LoadAnnotations'),
dict(type='PackSegInputs')
]

tta_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='LoadCityscapesDisparity'),
dict(type='ConcatRGBDispTo4Ch', keep_rgb_dtype=False, delete_disp=True),
dict(
type='TestTimeAug',
transforms=[
[
dict(type='Resize', scale_factor=r, keep_ratio=True)
for r in [0.5, 0.75, 1.0, 1.25, 1.5]
],
[
dict(type='RandomFlip', prob=0., direction='horizontal'),
dict(type='RandomFlip', prob=1., direction='horizontal')
],
[dict(type='LoadAnnotations')],
[dict(type='PackSegInputs')]
])
]

data_root = '/home/featurize/data/cityscapes'
train_dataloader = dict(
batch_size=4,
num_workers=4,
persistent_workers=True,
sampler=dict(type='InfiniteSampler', shuffle=True),
dataset=dict(
type='CityscapesDataset',
data_root=data_root,
data_prefix=dict(img_path='leftImg8bit/train', seg_map_path='gtFine/train'),
pipeline=train_pipeline))
val_dataloader = dict(
batch_size=1,
num_workers=4,
persistent_workers=True,
sampler=dict(type='DefaultSampler', shuffle=False),
dataset=dict(
type='CityscapesDataset',
data_root=data_root,
data_prefix=dict(img_path='leftImg8bit/val', seg_map_path='gtFine/val'),
pipeline=test_pipeline))
test_dataloader = val_dataloader

test_evaluator = dict(
type='IoUMetric',
iou_metrics=['mIoU', 'mDice', 'mFscore', 'mAcc', 'mPrecision', 'mRecall'],
)

val_evaluator = test_evaluator

optim_wrapper = dict(
type='AmpOptimWrapper',
optimizer=dict(type='AdamW', lr=6e-5, betas=(0.9, 0.999), weight_decay=0.01),
paramwise_cfg=dict(
custom_keys=dict(
pos_block=dict(decay_mult=0.),
norm=dict(decay_mult=0.),
head=dict(lr_mult=10.),
backbone_disp=dict(lr_mult=1.0))))

param_scheduler = [
dict(type='LinearLR', start_factor=1e-6, by_epoch=False, begin=0, end=1500),
dict(
type='PolyLR',
eta_min=0.0,
power=1.0,
begin=1500,
end=80000,
by_epoch=False)
]

train_cfg = dict(type='IterBasedTrainLoop', max_iters=80000, val_interval=4000)
val_cfg = dict(type='ValLoop')
test_cfg = dict(type='TestLoop')

default_hooks = dict(
timer=dict(type='IterTimerHook'),
logger=dict(type='LoggerHook', interval=50, log_metric_by_epoch=False),
param_scheduler=dict(type='ParamSchedulerHook'),
checkpoint=dict(type='CheckpointHook', by_epoch=False, interval=4000,save_best='mIoU', max_keep_ckpts=3),
sampler_seed=dict(type='DistSamplerSeedHook'),
visualization=dict(type='SegVisualizationHook'))

auto_scale_lr = dict(enable=False, base_batch_size=16)


work_dir = './work_dirs/ZCDataset-Segformer-20251129'
199 changes: 199 additions & 0 deletions configs/cm-dgseg/cm_dgseg_b0_cityscapes_ft.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,199 @@
# Copyright (c) OpenMMLab. All rights reserved.
"""CM-DGSeg B0 configuration for Cityscapes.

Run command:
python tools/train.py configs/cm-dgseg/cm_dgseg_b0_cityscapes.py
python tools/test.py configs/cm-dgseg/cm_dgseg_b0_cityscapes.py <checkpoint> --eval mIoU
"""

_base_ = [
'../_base_/datasets/cityscapes.py',
'../_base_/default_runtime.py',
]

crop_size = (512, 1024)
num_classes = 19
norm_cfg = dict(type='BN', requires_grad=True)
data_preprocessor = dict(
type='SegDataPreProcessor',
mean=[123.675, 116.28, 103.53, 0.0],
std=[58.395, 57.12, 57.375, 1.0],
bgr_to_rgb=True,
pad_val=0,
seg_pad_val=255,
size=crop_size)

model = dict(
type='DualStreamEncoderDecoder',
data_preprocessor=data_preprocessor,
backbone_rgb=dict(
type='MixVisionTransformer',
init_cfg=dict(type='Pretrained', checkpoint='https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segformer/mit_b0_20220624-7e0fe6dd.pth'),
in_channels=3,
embed_dims=32,
num_stages=4,
num_layers=[2, 2, 2, 2],
num_heads=[1, 2, 5, 8],
patch_sizes=[7, 3, 3, 3],
sr_ratios=[8, 4, 2, 1],
out_indices=(0, 1, 2, 3),
mlp_ratio=4,
qkv_bias=True,
drop_rate=0.0,
attn_drop_rate=0.0,
drop_path_rate=0.1),
backbone_disp=dict(
type='MixVisionTransformer',
in_channels=1,
embed_dims=16,
num_stages=4,
num_layers=[2, 2, 2, 2],
num_heads=[1, 1, 3, 4],
patch_sizes=[7, 3, 3, 3],
sr_ratios=[8, 4, 2, 1],
out_indices=(0, 1, 2, 3),
mlp_ratio=4,
qkv_bias=True,
drop_rate=0.0,
attn_drop_rate=0.0,
drop_path_rate=0.05),
decode_head=dict(
type='CMFSegFormerHead',
in_channels=[32, 64, 160, 256],
disp_in_channels=[16, 16, 48, 64],
in_index=[0, 1, 2, 3],
channels=256,
dropout_ratio=0.1,
num_classes=num_classes,
norm_cfg=norm_cfg,
align_corners=False,
with_boundary=False,
ignore_index=255,
loss_decode=[
dict(
type='CrossEntropyLoss',
use_sigmoid=False,
loss_weight=1.0,
avg_non_ignore=True
),
dict(
type='LovaszLoss',
per_image=False,
reduction='none',
loss_weight=0.4
)
]
),
train_cfg=dict(),
test_cfg=dict(mode='slide', crop_size=crop_size, stride=(768, 768)))

# Data pipelines: disable photometric distortion to keep disparity channel clean.
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='LoadCityscapesDisparity'),
dict(type='ConcatRGBDispTo4Ch', keep_rgb_dtype=False),
dict(type='LoadAnnotations'),
dict(type='RandomResize', scale=(2048, 1024), ratio_range=(0.5, 2.0), keep_ratio=True),
dict(type='RandomCrop', crop_size=crop_size, cat_max_ratio=0.75),
dict(type='RandomFlip', prob=0.5),
dict(type='PackSegInputs')
]

test_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='LoadCityscapesDisparity'),
dict(type='ConcatRGBDispTo4Ch', keep_rgb_dtype=False, delete_disp=True),
dict(type='Resize', scale=(2048, 1024), keep_ratio=True),
dict(type='LoadAnnotations'),
dict(type='PackSegInputs')
]

tta_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='LoadCityscapesDisparity'),
dict(type='ConcatRGBDispTo4Ch', keep_rgb_dtype=False, delete_disp=True),
dict(
type='TestTimeAug',
transforms=[
[
dict(type='Resize', scale_factor=r, keep_ratio=True)
for r in [0.5, 0.75, 1.0, 1.25, 1.5]
],
[
dict(type='RandomFlip', prob=0., direction='horizontal'),
dict(type='RandomFlip', prob=1., direction='horizontal')
],
[dict(type='LoadAnnotations')],
[dict(type='PackSegInputs')]
])
]

data_root = '/home/featurize/data/cityscapes'
train_dataloader = dict(
batch_size=4,
num_workers=4,
persistent_workers=True,
sampler=dict(type='InfiniteSampler', shuffle=True),
dataset=dict(
type='CityscapesDataset',
data_root=data_root,
data_prefix=dict(img_path='leftImg8bit/train', seg_map_path='gtFine/train'),
pipeline=train_pipeline))
val_dataloader = dict(
batch_size=1,
num_workers=4,
persistent_workers=True,
sampler=dict(type='DefaultSampler', shuffle=False),
dataset=dict(
type='CityscapesDataset',
data_root=data_root,
data_prefix=dict(img_path='leftImg8bit/val', seg_map_path='gtFine/val'),
pipeline=test_pipeline))
test_dataloader = val_dataloader

test_evaluator = dict(
type='IoUMetric',
iou_metrics=['mIoU', 'mDice', 'mFscore', 'mAcc', 'mPrecision', 'mRecall'],
)

val_evaluator = test_evaluator

optim_wrapper = dict(
type='AmpOptimWrapper',
optimizer=dict(type='AdamW', lr=2e-5, betas=(0.9, 0.999), weight_decay=0.01),
paramwise_cfg=dict(
custom_keys=dict(
pos_block=dict(decay_mult=0.),
norm=dict(decay_mult=0.),
head=dict(lr_mult=10.),
backbone_disp=dict(lr_mult=1.0))))

param_scheduler = [
dict(type='LinearLR', start_factor=1e-6, by_epoch=False, begin=0, end=1500),
dict(
type='PolyLR',
eta_min=0.0,
power=1.0,
begin=1500,
end=20000,
by_epoch=False)
]

train_cfg = dict(type='IterBasedTrainLoop', max_iters=20000, val_interval=2000)
val_cfg = dict(type='ValLoop')
test_cfg = dict(type='TestLoop')

default_hooks = dict(
timer=dict(type='IterTimerHook'),
logger=dict(type='LoggerHook', interval=50, log_metric_by_epoch=False),
param_scheduler=dict(type='ParamSchedulerHook'),
checkpoint=dict(type='CheckpointHook', by_epoch=False, interval=2000,save_best='mIoU', max_keep_ckpts=3),
sampler_seed=dict(type='DistSamplerSeedHook'),
visualization=dict(type='SegVisualizationHook'))

auto_scale_lr = dict(enable=False, base_batch_size=16)


load_from = './work_dirs/ZCDataset-Segformer-20251129/iter_80000.pth'

work_dir = './work_dirs/ZCDataset-Segformer-20251130-ft'
Loading