Skip to content

Mmseg1.x-updated #27

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

Mmseg1.x-updated #27

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
fe22a34
Migrated EasyPortrait Dataset classes to MMSegmentation1.x
guimCC Jun 27, 2024
08e92b0
Migrated Dataset configs to MMSegmentation 1.x
guimCC Jun 27, 2024
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
97 changes: 63 additions & 34 deletions pipelines/local_configs/__base__/datasets/easyportrait_1024x1024.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,59 +1,88 @@
# dataset settings
dataset_type = 'EasyPortraitDataset'
data_root = 'path/to/data/EasyPortrait'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
dataset_type = 'EasyPortraitFPDataset'
data_root = 'data/easyportrait/'
crop_size = (1024, 1024)

train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='LoadAnnotations'),
dict(type='Pad', size=(1920, 1920), pad_val=0, seg_pad_val=255),
dict(type='Resize', img_scale=(1024, 1024)),
dict(
type='RandomResize',
scale=(1024, 1024),
ratio_range=(1.0, 1.0), # keep the aspect ratio
keep_ratio=True),


dict(type='RandomCrop', crop_size=crop_size),
# We don't use RandomFlip, but need it in the code to fix error: https://github.com/open-mmlab/mmsegmentation/issues/231
dict(type='RandomFlip', prob=0.0),
dict(type='RandomFlip', prob=0.0),
dict(type='PhotoMetricDistortion',
brightness_delta=16,
contrast_range=(0.5, 1.0),
saturation_range=(0.5, 1.0),
hue_delta=9),
dict(type='Normalize', **img_norm_cfg),
dict(type='DefaultFormatBundle'),
dict(type='Collect', keys=['img', 'gt_semantic_seg']),
hue_delta=9),
dict(type='PackSegInputs')
]

test_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='Resize', scale=(1024, 1024), keep_ratio=True),
dict(type='LoadAnnotations'),
dict(type='PackSegInputs')

]

img_ratios = [0.5, 0.75, 1.0, 1.25, 1.5, 1.75]
tta_pipeline = [
dict(type='LoadImageFromFile', backend_args=None),
dict(
type='MultiScaleFlipAug',
img_scale=(1024, 1024),
flip=False,
type='TestTimeAug',
transforms=[
dict(type='Resize', keep_ratio=True),
dict(type='Normalize', **img_norm_cfg),
dict(type='ImageToTensor', keys=['img']),
dict(type='Collect', keys=['img']),
[
dict(type='Resize', scale_factor=r, keep_ratio=True)
for r in img_ratios
],
[
dict(type='RandomFlip', prob=0., direction='horizontal'),
dict(type='RandomFlip', prob=1., direction='horizontal')
], [dict(type='LoadAnnotations')], [dict(type='PackSegInputs')]
])
]

data = dict(
samples_per_gpu=4,
workers_per_gpu=4,
train=dict(
train_dataloader = dict(
# Increased batch size, maximmum on the
batch_size=4,
num_workers=4,
persistent_workers=True,
sampler=dict(type='InfiniteSampler', shuffle=True),
dataset=dict(
type=dataset_type,
data_root=data_root,
img_dir='images/train',
ann_dir='annotations/train',
pipeline=train_pipeline),
val=dict(
data_prefix=dict(
img_path='images/train', seg_map_path='annotations/train'),
pipeline=train_pipeline))

val_dataloader = dict(
batch_size=1,
num_workers=4,
persistent_workers=True,
sampler=dict(type='DefaultSampler', shuffle=False), # Do not shuffle during validation
dataset=dict(
type=dataset_type,
data_root=data_root,
img_dir='images/val',
ann_dir='annotations/val',
pipeline=test_pipeline),
test=dict(
data_prefix=dict(
img_path='images/val', seg_map_path='annotations/val'),
pipeline=test_pipeline))
test_dataloader = dict(
batch_size=1,
num_workers=4,
persistent_workers=True,
sampler=dict(type='DefaultSampler', shuffle=False), # Do not shuffle during validation
dataset=dict(
type=dataset_type,
data_root=data_root,
img_dir='images/test',
ann_dir='annotations/test',
pipeline=test_pipeline))
data_prefix=dict(
img_path='images/test', seg_map_path='annotations/test'),
pipeline=test_pipeline))

val_evaluator = dict(type='IoUMetric', iou_metrics=['mIoU']) # metric to measure the accuracy (mean IoU)
test_evaluator = val_evaluator
95 changes: 62 additions & 33 deletions pipelines/local_configs/__base__/datasets/easyportrait_384x384.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,59 +1,88 @@
# dataset settings
dataset_type = 'EasyPortraitDataset'
data_root = 'path/to/data/EasyPortrait'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
dataset_type = 'EasyPortraitFPDataset'
data_root = 'data/easyportrait/'
crop_size = (384, 384)

train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='LoadAnnotations'),
dict(type='Pad', size=(1920, 1920), pad_val=0, seg_pad_val=255),
dict(type='Resize', img_scale=(384, 384)),
dict(
type='RandomResize',
scale=(384, 384),
ratio_range=(1.0, 1.0), # keep the aspect ratio
keep_ratio=True),


dict(type='RandomCrop', crop_size=crop_size),
# We don't use RandomFlip, but need it in the code to fix error: https://github.com/open-mmlab/mmsegmentation/issues/231
dict(type='RandomFlip', prob=0.0),
dict(type='PhotoMetricDistortion',
brightness_delta=16,
contrast_range=(0.5, 1.0),
saturation_range=(0.5, 1.0),
hue_delta=9),
dict(type='Normalize', **img_norm_cfg),
dict(type='DefaultFormatBundle'),
dict(type='Collect', keys=['img', 'gt_semantic_seg']),
hue_delta=9),
dict(type='PackSegInputs')
]

test_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='Resize', scale=(384, 384), keep_ratio=True),
dict(type='LoadAnnotations'),
dict(type='PackSegInputs')

]

img_ratios = [0.5, 0.75, 1.0, 1.25, 1.5, 1.75]
tta_pipeline = [
dict(type='LoadImageFromFile', backend_args=None),
dict(
type='MultiScaleFlipAug',
img_scale=(384, 384),
flip=False,
type='TestTimeAug',
transforms=[
dict(type='Resize', keep_ratio=True),
dict(type='Normalize', **img_norm_cfg),
dict(type='ImageToTensor', keys=['img']),
dict(type='Collect', keys=['img']),
[
dict(type='Resize', scale_factor=r, keep_ratio=True)
for r in img_ratios
],
[
dict(type='RandomFlip', prob=0., direction='horizontal'),
dict(type='RandomFlip', prob=1., direction='horizontal')
], [dict(type='LoadAnnotations')], [dict(type='PackSegInputs')]
])
]

data = dict(
samples_per_gpu=4,
workers_per_gpu=4,
train=dict(
train_dataloader = dict(
# Increased batch size, maximmum on the
batch_size=4,
num_workers=4,
persistent_workers=True,
sampler=dict(type='InfiniteSampler', shuffle=True),
dataset=dict(
type=dataset_type,
data_root=data_root,
img_dir='images/train',
ann_dir='annotations/train',
pipeline=train_pipeline),
val=dict(
data_prefix=dict(
img_path='images/train', seg_map_path='annotations/train'),
pipeline=train_pipeline))

val_dataloader = dict(
batch_size=1,
num_workers=4,
persistent_workers=True,
sampler=dict(type='DefaultSampler', shuffle=False), # Do not shuffle during validation
dataset=dict(
type=dataset_type,
data_root=data_root,
img_dir='images/val',
ann_dir='annotations/val',
pipeline=test_pipeline),
test=dict(
data_prefix=dict(
img_path='images/val', seg_map_path='annotations/val'),
pipeline=test_pipeline))
test_dataloader = dict(
batch_size=1,
num_workers=4,
persistent_workers=True,
sampler=dict(type='DefaultSampler', shuffle=False), # Do not shuffle during validation
dataset=dict(
type=dataset_type,
data_root=data_root,
img_dir='images/test',
ann_dir='annotations/test',
pipeline=test_pipeline))
data_prefix=dict(
img_path='images/test', seg_map_path='annotations/test'),
pipeline=test_pipeline))

val_evaluator = dict(type='IoUMetric', iou_metrics=['mIoU']) # metric to measure the accuracy (mean IoU)
test_evaluator = val_evaluator
95 changes: 62 additions & 33 deletions pipelines/local_configs/__base__/datasets/easyportrait_512x512.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,59 +1,88 @@
# dataset settings
dataset_type = 'EasyPortraitDataset'
data_root = 'path/to/data/EasyPortrait'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
dataset_type = 'EasyPortraitFPDataset'
data_root = 'data/easyportrait/'
crop_size = (512, 512)

train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='LoadAnnotations'),
dict(type='Pad', size=(1920, 1920), pad_val=0, seg_pad_val=255),
dict(type='Resize', img_scale=(512, 512)),
dict(
type='RandomResize',
scale=(512, 512),
ratio_range=(1.0, 1.0), # keep the aspect ratio
keep_ratio=True),


dict(type='RandomCrop', crop_size=crop_size),
# We don't use RandomFlip, but need it in the code to fix error: https://github.com/open-mmlab/mmsegmentation/issues/231
dict(type='RandomFlip', prob=0.0),
dict(type='PhotoMetricDistortion',
brightness_delta=16,
contrast_range=(0.5, 1.0),
saturation_range=(0.5, 1.0),
hue_delta=9),
dict(type='Normalize', **img_norm_cfg),
dict(type='DefaultFormatBundle'),
dict(type='Collect', keys=['img', 'gt_semantic_seg']),
hue_delta=9),
dict(type='PackSegInputs')
]

test_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='Resize', scale=(512, 512), keep_ratio=True),
dict(type='LoadAnnotations'),
dict(type='PackSegInputs')

]

img_ratios = [0.5, 0.75, 1.0, 1.25, 1.5, 1.75]
tta_pipeline = [
dict(type='LoadImageFromFile', backend_args=None),
dict(
type='MultiScaleFlipAug',
img_scale=(512, 512),
flip=False,
type='TestTimeAug',
transforms=[
dict(type='Resize', keep_ratio=True),
dict(type='Normalize', **img_norm_cfg),
dict(type='ImageToTensor', keys=['img']),
dict(type='Collect', keys=['img']),
[
dict(type='Resize', scale_factor=r, keep_ratio=True)
for r in img_ratios
],
[
dict(type='RandomFlip', prob=0., direction='horizontal'),
dict(type='RandomFlip', prob=1., direction='horizontal')
], [dict(type='LoadAnnotations')], [dict(type='PackSegInputs')]
])
]

data = dict(
samples_per_gpu=4,
workers_per_gpu=4,
train=dict(
train_dataloader = dict(
# Increased batch size, maximmum on the
batch_size=4,
num_workers=4,
persistent_workers=True,
sampler=dict(type='InfiniteSampler', shuffle=True),
dataset=dict(
type=dataset_type,
data_root=data_root,
img_dir='images/train',
ann_dir='annotations/train',
pipeline=train_pipeline),
val=dict(
data_prefix=dict(
img_path='images/train', seg_map_path='annotations/train'),
pipeline=train_pipeline))

val_dataloader = dict(
batch_size=1,
num_workers=4,
persistent_workers=True,
sampler=dict(type='DefaultSampler', shuffle=False), # Do not shuffle during validation
dataset=dict(
type=dataset_type,
data_root=data_root,
img_dir='images/val',
ann_dir='annotations/val',
pipeline=test_pipeline),
test=dict(
data_prefix=dict(
img_path='images/val', seg_map_path='annotations/val'),
pipeline=test_pipeline))
test_dataloader = dict(
batch_size=1,
num_workers=4,
persistent_workers=True,
sampler=dict(type='DefaultSampler', shuffle=False), # Do not shuffle during validation
dataset=dict(
type=dataset_type,
data_root=data_root,
img_dir='images/test',
ann_dir='annotations/test',
pipeline=test_pipeline))
data_prefix=dict(
img_path='images/test', seg_map_path='annotations/test'),
pipeline=test_pipeline))

val_evaluator = dict(type='IoUMetric', iou_metrics=['mIoU']) # metric to measure the accuracy (mean IoU)
test_evaluator = val_evaluator
Loading