Skip to content

GitLab

Commit 37c8cebc authored by Sugon_ldc's avatar Sugon_ldc
Browse files

add new model

parents
Pipeline #318 failed with stages
in 0 seconds
Hide whitespace changes
Inline Side-by-side
Showing with 3032 additions and 0 deletions
configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/higherhrnet_w32_coco_512x512_udp.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
checkpoint_config = dict(interval=50)
evaluation = dict(interval=50, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=0.0015,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[200, 260])
total_epochs = 300
channel_cfg = dict(
dataset_joints=17,
dataset_channel=[
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
],
inference_channel=[
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
])
data_cfg = dict(
image_size=512,
base_size=256,
base_sigma=2,
heatmap_size=[128, 256],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
num_scales=2,
scale_aware_sigma=False,
)
# model settings
model = dict(
type='AssociativeEmbedding',
pretrained='https://download.openmmlab.com/mmpose/'
'pretrain_models/hrnet_w32-36af842e.pth',
backbone=dict(
type='HRNet',
in_channels=3,
extra=dict(
stage1=dict(
num_modules=1,
num_branches=1,
block='BOTTLENECK',
num_blocks=(4, ),
num_channels=(64, )),
stage2=dict(
num_modules=1,
num_branches=2,
block='BASIC',
num_blocks=(4, 4),
num_channels=(32, 64)),
stage3=dict(
num_modules=4,
num_branches=3,
block='BASIC',
num_blocks=(4, 4, 4),
num_channels=(32, 64, 128)),
stage4=dict(
num_modules=3,
num_branches=4,
block='BASIC',
num_blocks=(4, 4, 4, 4),
num_channels=(32, 64, 128, 256))),
),
keypoint_head=dict(
type='AEHigherResolutionHead',
in_channels=32,
num_joints=17,
tag_per_joint=True,
extra=dict(final_conv_kernel=1, ),
num_deconv_layers=1,
num_deconv_filters=[32],
num_deconv_kernels=[4],
num_basic_blocks=4,
cat_output=[True],
with_ae_loss=[True, False],
loss_keypoint=dict(
type='MultiLossFactory',
num_joints=17,
num_stages=2,
ae_loss_type='exp',
with_ae_loss=[True, False],
push_loss_factor=[0.001, 0.001],
pull_loss_factor=[0.001, 0.001],
with_heatmaps_loss=[True, True],
heatmaps_loss_factor=[1.0, 1.0])),
train_cfg=dict(),
test_cfg=dict(
num_joints=channel_cfg['dataset_joints'],
max_num_people=30,
scale_factor=[1],
with_heatmaps=[True, True],
with_ae=[True, False],
project2image=False,
align_corners=True,
nms_kernel=5,
nms_padding=2,
tag_per_joint=True,
detection_threshold=0.1,
tag_threshold=1,
use_detection_val=True,
ignore_too_much=False,
adjust=True,
refine=True,
flip_test=True,
use_udp=True))
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='BottomUpRandomAffine',
rot_factor=30,
scale_factor=[0.75, 1.5],
scale_type='short',
trans_factor=40,
use_udp=True),
dict(type='BottomUpRandomFlip', flip_prob=0.5),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(
type='BottomUpGenerateTarget',
sigma=2,
max_num_people=30,
use_udp=True,
),
dict(
type='Collect',
keys=['img', 'joints', 'targets', 'masks'],
meta_keys=[]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='BottomUpGetImgSize', test_scale_factor=[1], use_udp=True),
dict(
type='BottomUpResizeAlign',
transforms=[
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225])
],
use_udp=True),
dict(
type='Collect',
keys=['img'],
meta_keys=[
'image_file', 'aug_data', 'test_scale_factor', 'base_size',
'center', 'scale', 'flip_index'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
workers_per_gpu=2,
train_dataloader=dict(samples_per_gpu=24),
val_dataloader=dict(samples_per_gpu=1),
test_dataloader=dict(samples_per_gpu=1),
train=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
img_prefix=f'{data_root}/train2017/',
data_cfg=data_cfg,
pipeline=train_pipeline,
dataset_info={{_base_.dataset_info}}),
val=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=val_pipeline,
dataset_info={{_base_.dataset_info}}),
test=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=test_pipeline,
dataset_info={{_base_.dataset_info}}),
)
configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/higherhrnet_w32_coco_640x640.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
checkpoint_config = dict(interval=50)
evaluation = dict(interval=50, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=0.0015,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[200, 260])
total_epochs = 300
channel_cfg = dict(
dataset_joints=17,
dataset_channel=[
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
],
inference_channel=[
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
])
data_cfg = dict(
image_size=640,
base_size=320,
base_sigma=2,
heatmap_size=[160, 320],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
num_scales=2,
scale_aware_sigma=False,
)
# model settings
model = dict(
type='AssociativeEmbedding',
pretrained='https://download.openmmlab.com/mmpose/'
'pretrain_models/hrnet_w32-36af842e.pth',
backbone=dict(
type='HRNet',
in_channels=3,
extra=dict(
stage1=dict(
num_modules=1,
num_branches=1,
block='BOTTLENECK',
num_blocks=(4, ),
num_channels=(64, )),
stage2=dict(
num_modules=1,
num_branches=2,
block='BASIC',
num_blocks=(4, 4),
num_channels=(32, 64)),
stage3=dict(
num_modules=4,
num_branches=3,
block='BASIC',
num_blocks=(4, 4, 4),
num_channels=(32, 64, 128)),
stage4=dict(
num_modules=3,
num_branches=4,
block='BASIC',
num_blocks=(4, 4, 4, 4),
num_channels=(32, 64, 128, 256))),
),
keypoint_head=dict(
type='AEHigherResolutionHead',
in_channels=32,
num_joints=17,
tag_per_joint=True,
extra=dict(final_conv_kernel=1, ),
num_deconv_layers=1,
num_deconv_filters=[32],
num_deconv_kernels=[4],
num_basic_blocks=4,
cat_output=[True],
with_ae_loss=[True, False],
loss_keypoint=dict(
type='MultiLossFactory',
num_joints=17,
num_stages=2,
ae_loss_type='exp',
with_ae_loss=[True, False],
push_loss_factor=[0.001, 0.001],
pull_loss_factor=[0.001, 0.001],
with_heatmaps_loss=[True, True],
heatmaps_loss_factor=[1.0, 1.0])),
train_cfg=dict(),
test_cfg=dict(
num_joints=channel_cfg['dataset_joints'],
max_num_people=30,
scale_factor=[1],
with_heatmaps=[True, True],
with_ae=[True, False],
project2image=True,
align_corners=False,
nms_kernel=5,
nms_padding=2,
tag_per_joint=True,
detection_threshold=0.1,
tag_threshold=1,
use_detection_val=True,
ignore_too_much=False,
adjust=True,
refine=True,
flip_test=True))
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='BottomUpRandomAffine',
rot_factor=30,
scale_factor=[0.75, 1.5],
scale_type='short',
trans_factor=40),
dict(type='BottomUpRandomFlip', flip_prob=0.5),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(
type='BottomUpGenerateTarget',
sigma=2,
max_num_people=30,
),
dict(
type='Collect',
keys=['img', 'joints', 'targets', 'masks'],
meta_keys=[]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='BottomUpGetImgSize', test_scale_factor=[1]),
dict(
type='BottomUpResizeAlign',
transforms=[
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
]),
dict(
type='Collect',
keys=['img'],
meta_keys=[
'image_file', 'aug_data', 'test_scale_factor', 'base_size',
'center', 'scale', 'flip_index'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
workers_per_gpu=2,
train_dataloader=dict(samples_per_gpu=16),
val_dataloader=dict(samples_per_gpu=1),
test_dataloader=dict(samples_per_gpu=1),
train=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
img_prefix=f'{data_root}/train2017/',
data_cfg=data_cfg,
pipeline=train_pipeline,
dataset_info={{_base_.dataset_info}}),
val=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=val_pipeline,
dataset_info={{_base_.dataset_info}}),
test=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=test_pipeline,
dataset_info={{_base_.dataset_info}}),
)
configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/higherhrnet_w32_coco_640x640_udp.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
checkpoint_config = dict(interval=50)
evaluation = dict(interval=50, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=0.0015,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[200, 260])
total_epochs = 300
channel_cfg = dict(
dataset_joints=17,
dataset_channel=[
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
],
inference_channel=[
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
])
data_cfg = dict(
image_size=640,
base_size=320,
base_sigma=2,
heatmap_size=[160, 320],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
num_scales=2,
scale_aware_sigma=False,
)
# model settings
model = dict(
type='AssociativeEmbedding',
pretrained='https://download.openmmlab.com/mmpose/'
'pretrain_models/hrnet_w32-36af842e.pth',
backbone=dict(
type='HRNet',
in_channels=3,
extra=dict(
stage1=dict(
num_modules=1,
num_branches=1,
block='BOTTLENECK',
num_blocks=(4, ),
num_channels=(64, )),
stage2=dict(
num_modules=1,
num_branches=2,
block='BASIC',
num_blocks=(4, 4),
num_channels=(32, 64)),
stage3=dict(
num_modules=4,
num_branches=3,
block='BASIC',
num_blocks=(4, 4, 4),
num_channels=(32, 64, 128)),
stage4=dict(
num_modules=3,
num_branches=4,
block='BASIC',
num_blocks=(4, 4, 4, 4),
num_channels=(32, 64, 128, 256))),
),
keypoint_head=dict(
type='AEHigherResolutionHead',
in_channels=32,
num_joints=17,
tag_per_joint=True,
extra=dict(final_conv_kernel=1, ),
num_deconv_layers=1,
num_deconv_filters=[32],
num_deconv_kernels=[4],
num_basic_blocks=4,
cat_output=[True],
with_ae_loss=[True, False],
loss_keypoint=dict(
type='MultiLossFactory',
num_joints=17,
num_stages=2,
ae_loss_type='exp',
with_ae_loss=[True, False],
push_loss_factor=[0.001, 0.001],
pull_loss_factor=[0.001, 0.001],
with_heatmaps_loss=[True, True],
heatmaps_loss_factor=[1.0, 1.0])),
train_cfg=dict(),
test_cfg=dict(
num_joints=channel_cfg['dataset_joints'],
max_num_people=30,
scale_factor=[1],
with_heatmaps=[True, True],
with_ae=[True, False],
project2image=False,
align_corners=True,
nms_kernel=5,
nms_padding=2,
tag_per_joint=True,
detection_threshold=0.1,
tag_threshold=1,
use_detection_val=True,
ignore_too_much=False,
adjust=True,
refine=True,
flip_test=True,
use_udp=True))
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='BottomUpRandomAffine',
rot_factor=30,
scale_factor=[0.75, 1.5],
scale_type='short',
trans_factor=40,
use_udp=True),
dict(type='BottomUpRandomFlip', flip_prob=0.5),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(
type='BottomUpGenerateTarget',
sigma=2,
max_num_people=30,
use_udp=True,
),
dict(
type='Collect',
keys=['img', 'joints', 'targets', 'masks'],
meta_keys=[]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='BottomUpGetImgSize', test_scale_factor=[1], use_udp=True),
dict(
type='BottomUpResizeAlign',
transforms=[
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225])
],
use_udp=True),
dict(
type='Collect',
keys=['img'],
meta_keys=[
'image_file', 'aug_data', 'test_scale_factor', 'base_size',
'center', 'scale', 'flip_index'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
workers_per_gpu=2,
train_dataloader=dict(samples_per_gpu=16),
val_dataloader=dict(samples_per_gpu=1),
test_dataloader=dict(samples_per_gpu=1),
train=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
img_prefix=f'{data_root}/train2017/',
data_cfg=data_cfg,
pipeline=train_pipeline,
dataset_info={{_base_.dataset_info}}),
val=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=val_pipeline,
dataset_info={{_base_.dataset_info}}),
test=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=test_pipeline,
dataset_info={{_base_.dataset_info}}),
)
configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/higherhrnet_w48_coco_512x512.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
checkpoint_config = dict(interval=50)
evaluation = dict(interval=50, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=0.0015,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[200, 260])
total_epochs = 300
channel_cfg = dict(
dataset_joints=17,
dataset_channel=[
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
],
inference_channel=[
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
])
data_cfg = dict(
image_size=512,
base_size=256,
base_sigma=2,
heatmap_size=[128, 256],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
num_scales=2,
scale_aware_sigma=False,
)
# model settings
model = dict(
type='AssociativeEmbedding',
pretrained='https://download.openmmlab.com/mmpose/'
'pretrain_models/hrnet_w48-8ef0771d.pth',
backbone=dict(
type='HRNet',
in_channels=3,
extra=dict(
stage1=dict(
num_modules=1,
num_branches=1,
block='BOTTLENECK',
num_blocks=(4, ),
num_channels=(64, )),
stage2=dict(
num_modules=1,
num_branches=2,
block='BASIC',
num_blocks=(4, 4),
num_channels=(48, 96)),
stage3=dict(
num_modules=4,
num_branches=3,
block='BASIC',
num_blocks=(4, 4, 4),
num_channels=(48, 96, 192)),
stage4=dict(
num_modules=3,
num_branches=4,
block='BASIC',
num_blocks=(4, 4, 4, 4),
num_channels=(48, 96, 192, 384))),
),
keypoint_head=dict(
type='AEHigherResolutionHead',
in_channels=48,
num_joints=17,
tag_per_joint=True,
extra=dict(final_conv_kernel=1, ),
num_deconv_layers=1,
num_deconv_filters=[48],
num_deconv_kernels=[4],
num_basic_blocks=4,
cat_output=[True],
with_ae_loss=[True, False],
loss_keypoint=dict(
type='MultiLossFactory',
num_joints=17,
num_stages=2,
ae_loss_type='exp',
with_ae_loss=[True, False],
push_loss_factor=[0.001, 0.001],
pull_loss_factor=[0.001, 0.001],
with_heatmaps_loss=[True, True],
heatmaps_loss_factor=[1.0, 1.0])),
train_cfg=dict(),
test_cfg=dict(
num_joints=channel_cfg['dataset_joints'],
max_num_people=30,
scale_factor=[1],
with_heatmaps=[True, True],
with_ae=[True, False],
project2image=True,
align_corners=False,
nms_kernel=5,
nms_padding=2,
tag_per_joint=True,
detection_threshold=0.1,
tag_threshold=1,
use_detection_val=True,
ignore_too_much=False,
adjust=True,
refine=True,
flip_test=True))
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='BottomUpRandomAffine',
rot_factor=30,
scale_factor=[0.75, 1.5],
scale_type='short',
trans_factor=40),
dict(type='BottomUpRandomFlip', flip_prob=0.5),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(
type='BottomUpGenerateTarget',
sigma=2,
max_num_people=30,
),
dict(
type='Collect',
keys=['img', 'joints', 'targets', 'masks'],
meta_keys=[]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='BottomUpGetImgSize', test_scale_factor=[1]),
dict(
type='BottomUpResizeAlign',
transforms=[
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
]),
dict(
type='Collect',
keys=['img'],
meta_keys=[
'image_file', 'aug_data', 'test_scale_factor', 'base_size',
'center', 'scale', 'flip_index'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
workers_per_gpu=2,
train_dataloader=dict(samples_per_gpu=16),
val_dataloader=dict(samples_per_gpu=1),
test_dataloader=dict(samples_per_gpu=1),
train=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
img_prefix=f'{data_root}/train2017/',
data_cfg=data_cfg,
pipeline=train_pipeline,
dataset_info={{_base_.dataset_info}}),
val=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=val_pipeline,
dataset_info={{_base_.dataset_info}}),
test=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=test_pipeline,
dataset_info={{_base_.dataset_info}}),
)
configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/higherhrnet_w48_coco_512x512_udp.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
checkpoint_config = dict(interval=50)
evaluation = dict(interval=50, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=0.0015,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[200, 260])
total_epochs = 300
channel_cfg = dict(
dataset_joints=17,
dataset_channel=[
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
],
inference_channel=[
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
])
data_cfg = dict(
image_size=512,
base_size=256,
base_sigma=2,
heatmap_size=[128, 256],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
num_scales=2,
scale_aware_sigma=False,
)
# model settings
model = dict(
type='AssociativeEmbedding',
pretrained='https://download.openmmlab.com/mmpose/'
'pretrain_models/hrnet_w48-8ef0771d.pth',
backbone=dict(
type='HRNet',
in_channels=3,
extra=dict(
stage1=dict(
num_modules=1,
num_branches=1,
block='BOTTLENECK',
num_blocks=(4, ),
num_channels=(64, )),
stage2=dict(
num_modules=1,
num_branches=2,
block='BASIC',
num_blocks=(4, 4),
num_channels=(48, 96)),
stage3=dict(
num_modules=4,
num_branches=3,
block='BASIC',
num_blocks=(4, 4, 4),
num_channels=(48, 96, 192)),
stage4=dict(
num_modules=3,
num_branches=4,
block='BASIC',
num_blocks=(4, 4, 4, 4),
num_channels=(48, 96, 192, 384))),
),
keypoint_head=dict(
type='AEHigherResolutionHead',
in_channels=48,
num_joints=17,
tag_per_joint=True,
extra=dict(final_conv_kernel=1, ),
num_deconv_layers=1,
num_deconv_filters=[48],
num_deconv_kernels=[4],
num_basic_blocks=4,
cat_output=[True],
with_ae_loss=[True, False],
loss_keypoint=dict(
type='MultiLossFactory',
num_joints=17,
num_stages=2,
ae_loss_type='exp',
with_ae_loss=[True, False],
push_loss_factor=[0.001, 0.001],
pull_loss_factor=[0.001, 0.001],
with_heatmaps_loss=[True, True],
heatmaps_loss_factor=[1.0, 1.0])),
train_cfg=dict(),
test_cfg=dict(
num_joints=channel_cfg['dataset_joints'],
max_num_people=30,
scale_factor=[1],
with_heatmaps=[True, True],
with_ae=[True, False],
project2image=False,
align_corners=True,
nms_kernel=5,
nms_padding=2,
tag_per_joint=True,
detection_threshold=0.1,
tag_threshold=1,
use_detection_val=True,
ignore_too_much=False,
adjust=True,
refine=True,
flip_test=True,
use_udp=True))
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='BottomUpRandomAffine',
rot_factor=30,
scale_factor=[0.75, 1.5],
scale_type='short',
trans_factor=40,
use_udp=True),
dict(type='BottomUpRandomFlip', flip_prob=0.5),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(
type='BottomUpGenerateTarget',
sigma=2,
max_num_people=30,
use_udp=True,
),
dict(
type='Collect',
keys=['img', 'joints', 'targets', 'masks'],
meta_keys=[]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='BottomUpGetImgSize', test_scale_factor=[1], use_udp=True),
dict(
type='BottomUpResizeAlign',
transforms=[
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225])
],
use_udp=True),
dict(
type='Collect',
keys=['img'],
meta_keys=[
'image_file', 'aug_data', 'test_scale_factor', 'base_size',
'center', 'scale', 'flip_index'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
workers_per_gpu=2,
train_dataloader=dict(samples_per_gpu=16),
val_dataloader=dict(samples_per_gpu=1),
test_dataloader=dict(samples_per_gpu=1),
train=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
img_prefix=f'{data_root}/train2017/',
data_cfg=data_cfg,
pipeline=train_pipeline,
dataset_info={{_base_.dataset_info}}),
val=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=val_pipeline,
dataset_info={{_base_.dataset_info}}),
test=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=test_pipeline,
dataset_info={{_base_.dataset_info}}),
)
configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hourglass_ae_coco.md 0 → 100755
View file @ 37c8cebc
<!-- [ALGORITHM] -->
<details>
<summary align="right"><a href="https://arxiv.org/abs/1611.05424">Associative Embedding (NIPS'2017)</a></summary>
```bibtex
@inproceedings{newell2017associative,
title={Associative embedding: End-to-end learning for joint detection and grouping},
author={Newell, Alejandro and Huang, Zhiao and Deng, Jia},
booktitle={Advances in neural information processing systems},
pages={2277--2287},
year={2017}
}
```
</details>
<!-- [BACKBONE] -->
<details>
<summary align="right"><a href="https://arxiv.org/abs/1611.05424">HourglassAENet (NIPS'2017)</a></summary>
```bibtex
@inproceedings{newell2017associative,
title={Associative embedding: End-to-end learning for joint detection and grouping},
author={Newell, Alejandro and Huang, Zhiao and Deng, Jia},
booktitle={Advances in neural information processing systems},
pages={2277--2287},
year={2017}
}
```
</details>
<!-- [DATASET] -->
<details>
<summary align="right"><a href="https://link.springer.com/chapter/10.1007/978-3-319-10602-1_48">COCO (ECCV'2014)</a></summary>
```bibtex
@inproceedings{lin2014microsoft,
title={Microsoft coco: Common objects in context},
author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
booktitle={European conference on computer vision},
pages={740--755},
year={2014},
organization={Springer}
}
```
</details>
Results on COCO val2017 without multi-scale test
| Arch | Input Size | AP | AP<sup>50</sup> | AP<sup>75</sup> | AR | AR<sup>50</sup> | ckpt | log |
| :-------------------------------------------- | :--------: | :---: | :-------------: | :-------------: | :---: | :-------------: | :-------------------------------------------: | :-------------------------------------------: |
| [pose_hourglass_ae](/configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hourglass_ae_coco_512x512.py) | 512x512 | 0.613 | 0.833 | 0.667 | 0.659 | 0.850 | [ckpt](https://download.openmmlab.com/mmpose/bottom_up/hourglass_ae/hourglass_ae_coco_512x512-90af499f_20210920.pth) | [log](https://download.openmmlab.com/mmpose/bottom_up/hourglass_ae/hourglass_ae_coco_512x512_20210920.log.json) |
Results on COCO val2017 with multi-scale test. 3 default scales (\[2, 1, 0.5\]) are used
| Arch | Input Size | AP | AP<sup>50</sup> | AP<sup>75</sup> | AR | AR<sup>50</sup> | ckpt | log |
| :-------------------------------------------- | :--------: | :---: | :-------------: | :-------------: | :---: | :-------------: | :-------------------------------------------: | :-------------------------------------------: |
| [pose_hourglass_ae](/configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hourglass_ae_coco_512x512.py) | 512x512 | 0.667 | 0.855 | 0.723 | 0.707 | 0.877 | [ckpt](https://download.openmmlab.com/mmpose/bottom_up/hourglass_ae/hourglass_ae_coco_512x512-90af499f_20210920.pth) | [log](https://download.openmmlab.com/mmpose/bottom_up/hourglass_ae/hourglass_ae_coco_512x512_20210920.log.json) |
configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hourglass_ae_coco.yml 0 → 100755
View file @ 37c8cebc
Collections:
- Name: Associative Embedding
Paper:
Title: 'Associative embedding: End-to-end learning for joint detection and grouping'
URL: https://arxiv.org/abs/1611.05424
README: https://github.com/open-mmlab/mmpose/blob/master/docs/en/papers/algorithms/associative_embedding.md
Models:
- Config: configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hourglass_ae_coco_512x512.py
In Collection: Associative Embedding
Metadata:
Architecture: &id001
- Associative Embedding
- HourglassAENet
Training Data: COCO
Name: associative_embedding_hourglass_ae_coco_512x512
Results:
- Dataset: COCO
Metrics:
AP: 0.613
AP@0.5: 0.833
AP@0.75: 0.667
AR: 0.659
AR@0.5: 0.85
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/bottom_up/hourglass_ae/hourglass_ae_coco_512x512-90af499f_20210920.pth
- Config: configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hourglass_ae_coco_512x512.py
In Collection: Associative Embedding
Metadata:
Architecture: *id001
Training Data: COCO
Name: associative_embedding_hourglass_ae_coco_512x512
Results:
- Dataset: COCO
Metrics:
AP: 0.667
AP@0.5: 0.855
AP@0.75: 0.723
AR: 0.707
AR@0.5: 0.877
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/bottom_up/hourglass_ae/hourglass_ae_coco_512x512-90af499f_20210920.pth
configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hourglass_ae_coco_512x512.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
checkpoint_config = dict(interval=50)
evaluation = dict(interval=50, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=0.0015,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[200, 260])
total_epochs = 300
channel_cfg = dict(
dataset_joints=17,
dataset_channel=[
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
],
inference_channel=[
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
])
data_cfg = dict(
image_size=512,
base_size=256,
base_sigma=2,
heatmap_size=[128],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
num_scales=1,
scale_aware_sigma=False,
)
# model settings
model = dict(
type='AssociativeEmbedding',
pretrained=None,
backbone=dict(
type='HourglassAENet',
num_stacks=4,
out_channels=34,
),
keypoint_head=dict(
type='AEMultiStageHead',
in_channels=34,
out_channels=34,
num_stages=4,
num_deconv_layers=0,
extra=dict(final_conv_kernel=0),
loss_keypoint=dict(
type='MultiLossFactory',
num_joints=17,
num_stages=4,
ae_loss_type='exp',
with_heatmaps_loss=[True, True, True, True],
with_ae_loss=[True, True, True, True],
push_loss_factor=[0.001, 0.001, 0.001, 0.001],
pull_loss_factor=[0.001, 0.001, 0.001, 0.001],
heatmaps_loss_factor=[1.0, 1.0, 1.0, 1.0])),
train_cfg=dict(
num_joints=channel_cfg['dataset_joints'],
img_size=data_cfg['image_size']),
test_cfg=dict(
num_joints=channel_cfg['dataset_joints'],
max_num_people=30,
scale_factor=[1],
with_heatmaps=[True, True, True, True],
with_ae=[True, True, True, True],
select_output_index=[3],
project2image=True,
nms_kernel=5,
nms_padding=2,
tag_per_joint=True,
detection_threshold=0.1,
tag_threshold=1,
use_detection_val=True,
ignore_too_much=False,
adjust=True,
refine=True,
flip_test=True))
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='BottomUpRandomAffine',
rot_factor=30,
scale_factor=[0.75, 1.5],
scale_type='short',
trans_factor=40),
dict(type='BottomUpRandomFlip', flip_prob=0.5),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(
type='MultitaskGatherTarget',
pipeline_list=[
[dict(type='BottomUpGenerateTarget', sigma=2, max_num_people=30)],
],
pipeline_indices=[0] * 4,
keys=['targets', 'masks', 'joints']),
dict(
type='Collect',
keys=['img', 'joints', 'targets', 'masks'],
meta_keys=[]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='BottomUpGetImgSize', test_scale_factor=[1]),
dict(
type='BottomUpResizeAlign',
transforms=[
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
]),
dict(
type='Collect',
keys=['img'],
meta_keys=[
'image_file', 'aug_data', 'test_scale_factor', 'base_size',
'center', 'scale', 'flip_index'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
workers_per_gpu=2,
train_dataloader=dict(samples_per_gpu=6),
val_dataloader=dict(samples_per_gpu=1),
test_dataloader=dict(samples_per_gpu=1),
train=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
img_prefix=f'{data_root}/train2017/',
data_cfg=data_cfg,
pipeline=train_pipeline,
dataset_info={{_base_.dataset_info}}),
val=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=val_pipeline,
dataset_info={{_base_.dataset_info}}),
test=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=test_pipeline,
dataset_info={{_base_.dataset_info}}),
)
configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_coco.md 0 → 100755
View file @ 37c8cebc
<!-- [ALGORITHM] -->
<details>
<summary align="right"><a href="https://arxiv.org/abs/1611.05424">Associative Embedding (NIPS'2017)</a></summary>
```bibtex
@inproceedings{newell2017associative,
title={Associative embedding: End-to-end learning for joint detection and grouping},
author={Newell, Alejandro and Huang, Zhiao and Deng, Jia},
booktitle={Advances in neural information processing systems},
pages={2277--2287},
year={2017}
}
```
</details>
<!-- [ALGORITHM] -->
<details>
<summary align="right"><a href="http://openaccess.thecvf.com/content_CVPR_2019/html/Sun_Deep_High-Resolution_Representation_Learning_for_Human_Pose_Estimation_CVPR_2019_paper.html">HRNet (CVPR'2019)</a></summary>
```bibtex
@inproceedings{sun2019deep,
title={Deep high-resolution representation learning for human pose estimation},
author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
pages={5693--5703},
year={2019}
}
```
</details>
<!-- [DATASET] -->
<details>
<summary align="right"><a href="https://link.springer.com/chapter/10.1007/978-3-319-10602-1_48">COCO (ECCV'2014)</a></summary>
```bibtex
@inproceedings{lin2014microsoft,
title={Microsoft coco: Common objects in context},
author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
booktitle={European conference on computer vision},
pages={740--755},
year={2014},
organization={Springer}
}
```
</details>
Results on COCO val2017 without multi-scale test
| Arch | Input Size | AP | AP<sup>50</sup> | AP<sup>75</sup> | AR | AR<sup>50</sup> | ckpt | log |
| :-------------------------------------------- | :--------: | :---: | :-------------: | :-------------: | :---: | :-------------: | :-------------------------------------------: | :-------------------------------------------: |
| [HRNet-w32](/configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_w32_coco_512x512.py) | 512x512 | 0.654 | 0.863 | 0.720 | 0.710 | 0.892 | [ckpt](https://download.openmmlab.com/mmpose/bottom_up/hrnet_w32_coco_512x512-bcb8c247_20200816.pth) | [log](https://download.openmmlab.com/mmpose/bottom_up/hrnet_w32_coco_512x512_20200816.log.json) |
| [HRNet-w48](/configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_w48_coco_512x512.py) | 512x512 | 0.665 | 0.860 | 0.727 | 0.716 | 0.889 | [ckpt](https://download.openmmlab.com/mmpose/bottom_up/hrnet_w48_coco_512x512-cf72fcdf_20200816.pth) | [log](https://download.openmmlab.com/mmpose/bottom_up/hrnet_w48_coco_512x512_20200816.log.json) |
Results on COCO val2017 with multi-scale test. 3 default scales (\[2, 1, 0.5\]) are used
| Arch | Input Size | AP | AP<sup>50</sup> | AP<sup>75</sup> | AR | AR<sup>50</sup> | ckpt | log |
| :-------------------------------------------- | :--------: | :---: | :-------------: | :-------------: | :---: | :-------------: | :-------------------------------------------: | :-------------------------------------------: |
| [HRNet-w32](/configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_w32_coco_512x512.py) | 512x512 | 0.698 | 0.877 | 0.760 | 0.748 | 0.907 | [ckpt](https://download.openmmlab.com/mmpose/bottom_up/hrnet_w32_coco_512x512-bcb8c247_20200816.pth) | [log](https://download.openmmlab.com/mmpose/bottom_up/hrnet_w32_coco_512x512_20200816.log.json) |
| [HRNet-w48](/configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_w48_coco_512x512.py) | 512x512 | 0.712 | 0.880 | 0.771 | 0.757 | 0.909 | [ckpt](https://download.openmmlab.com/mmpose/bottom_up/hrnet_w48_coco_512x512-cf72fcdf_20200816.pth) | [log](https://download.openmmlab.com/mmpose/bottom_up/hrnet_w48_coco_512x512_20200816.log.json) |
configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_coco.yml 0 → 100755
View file @ 37c8cebc
Collections:
- Name: HRNet
Paper:
Title: Deep high-resolution representation learning for human pose estimation
URL: http://openaccess.thecvf.com/content_CVPR_2019/html/Sun_Deep_High-Resolution_Representation_Learning_for_Human_Pose_Estimation_CVPR_2019_paper.html
README: https://github.com/open-mmlab/mmpose/blob/master/docs/en/papers/backbones/hrnet.md
Models:
- Config: configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_w32_coco_512x512.py
In Collection: HRNet
Metadata:
Architecture: &id001
- Associative Embedding
- HRNet
Training Data: COCO
Name: associative_embedding_hrnet_w32_coco_512x512
Results:
- Dataset: COCO
Metrics:
AP: 0.654
AP@0.5: 0.863
AP@0.75: 0.72
AR: 0.71
AR@0.5: 0.892
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/bottom_up/hrnet_w32_coco_512x512-bcb8c247_20200816.pth
- Config: configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_w48_coco_512x512.py
In Collection: HRNet
Metadata:
Architecture: *id001
Training Data: COCO
Name: associative_embedding_hrnet_w48_coco_512x512
Results:
- Dataset: COCO
Metrics:
AP: 0.665
AP@0.5: 0.86
AP@0.75: 0.727
AR: 0.716
AR@0.5: 0.889
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/bottom_up/hrnet_w48_coco_512x512-cf72fcdf_20200816.pth
- Config: configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_w32_coco_512x512.py
In Collection: HRNet
Metadata:
Architecture: *id001
Training Data: COCO
Name: associative_embedding_hrnet_w32_coco_512x512
Results:
- Dataset: COCO
Metrics:
AP: 0.698
AP@0.5: 0.877
AP@0.75: 0.76
AR: 0.748
AR@0.5: 0.907
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/bottom_up/hrnet_w32_coco_512x512-bcb8c247_20200816.pth
- Config: configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_w48_coco_512x512.py
In Collection: HRNet
Metadata:
Architecture: *id001
Training Data: COCO
Name: associative_embedding_hrnet_w48_coco_512x512
Results:
- Dataset: COCO
Metrics:
AP: 0.712
AP@0.5: 0.88
AP@0.75: 0.771
AR: 0.757
AR@0.5: 0.909
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/bottom_up/hrnet_w48_coco_512x512-cf72fcdf_20200816.pth
configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_udp_coco.md 0 → 100755
View file @ 37c8cebc
<!-- [ALGORITHM] -->
<details>
<summary align="right"><a href="https://arxiv.org/abs/1611.05424">Associative Embedding (NIPS'2017)</a></summary>
```bibtex
@inproceedings{newell2017associative,
title={Associative embedding: End-to-end learning for joint detection and grouping},
author={Newell, Alejandro and Huang, Zhiao and Deng, Jia},
booktitle={Advances in neural information processing systems},
pages={2277--2287},
year={2017}
}
```
</details>
<!-- [BACKBONE] -->
<details>
<summary align="right"><a href="http://openaccess.thecvf.com/content_CVPR_2019/html/Sun_Deep_High-Resolution_Representation_Learning_for_Human_Pose_Estimation_CVPR_2019_paper.html">HRNet (CVPR'2019)</a></summary>
```bibtex
@inproceedings{sun2019deep,
title={Deep high-resolution representation learning for human pose estimation},
author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
pages={5693--5703},
year={2019}
}
```
</details>
<!-- [ALGORITHM] -->
<details>
<summary align="right"><a href="http://openaccess.thecvf.com/content_CVPR_2020/html/Huang_The_Devil_Is_in_the_Details_Delving_Into_Unbiased_Data_CVPR_2020_paper.html">UDP (CVPR'2020)</a></summary>
```bibtex
@InProceedings{Huang_2020_CVPR,
author = {Huang, Junjie and Zhu, Zheng and Guo, Feng and Huang, Guan},
title = {The Devil Is in the Details: Delving Into Unbiased Data Processing for Human Pose Estimation},
booktitle = {The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2020}
}
```
</details>
<!-- [DATASET] -->
<details>
<summary align="right"><a href="https://link.springer.com/chapter/10.1007/978-3-319-10602-1_48">COCO (ECCV'2014)</a></summary>
```bibtex
@inproceedings{lin2014microsoft,
title={Microsoft coco: Common objects in context},
author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
booktitle={European conference on computer vision},
pages={740--755},
year={2014},
organization={Springer}
}
```
</details>
Results on COCO val2017 without multi-scale test
| Arch | Input Size | AP | AP<sup>50</sup> | AP<sup>75</sup> | AR | AR<sup>50</sup> | ckpt | log |
| :-------------------------------------------- | :--------: | :---: | :-------------: | :-------------: | :---: | :-------------: | :-------------------------------------------: | :-------------------------------------------: |
| [HRNet-w32_udp](/configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_w32_coco_512x512_udp.py) | 512x512 | 0.671 | 0.863 | 0.729 | 0.717 | 0.889 | [ckpt](https://download.openmmlab.com/mmpose/bottom_up/hrnet_w32_coco_512x512_udp-91663bf9_20210220.pth) | [log](https://download.openmmlab.com/mmpose/bottom_up/hrnet_w32_coco_512x512_udp_20210220.log.json) |
| [HRNet-w48_udp](/configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_w48_coco_512x512_udp.py) | 512x512 | 0.681 | 0.872 | 0.741 | 0.725 | 0.892 | [ckpt](https://download.openmmlab.com/mmpose/bottom_up/hrnet_w48_coco_512x512_udp-de08fd8c_20210222.pth) | [log](https://download.openmmlab.com/mmpose/bottom_up/hrnet_w48_coco_512x512_udp_20210222.log.json) |
configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_udp_coco.yml 0 → 100755
View file @ 37c8cebc
Collections:
- Name: UDP
Paper:
Title: 'The Devil Is in the Details: Delving Into Unbiased Data Processing for
Human Pose Estimation'
URL: http://openaccess.thecvf.com/content_CVPR_2020/html/Huang_The_Devil_Is_in_the_Details_Delving_Into_Unbiased_Data_CVPR_2020_paper.html
README: https://github.com/open-mmlab/mmpose/blob/master/docs/en/papers/techniques/udp.md
Models:
- Config: configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_w32_coco_512x512_udp.py
In Collection: UDP
Metadata:
Architecture: &id001
- Associative Embedding
- HRNet
- UDP
Training Data: COCO
Name: associative_embedding_hrnet_w32_coco_512x512_udp
Results:
- Dataset: COCO
Metrics:
AP: 0.671
AP@0.5: 0.863
AP@0.75: 0.729
AR: 0.717
AR@0.5: 0.889
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/bottom_up/hrnet_w32_coco_512x512_udp-91663bf9_20210220.pth
- Config: configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_w48_coco_512x512_udp.py
In Collection: UDP
Metadata:
Architecture: *id001
Training Data: COCO
Name: associative_embedding_hrnet_w48_coco_512x512_udp
Results:
- Dataset: COCO
Metrics:
AP: 0.681
AP@0.5: 0.872
AP@0.75: 0.741
AR: 0.725
AR@0.5: 0.892
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/bottom_up/hrnet_w48_coco_512x512_udp-de08fd8c_20210222.pth
configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_w32_coco_512x512.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
checkpoint_config = dict(interval=50)
evaluation = dict(interval=50, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=0.0015,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[200, 260])
total_epochs = 300
channel_cfg = dict(
dataset_joints=17,
dataset_channel=[
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
],
inference_channel=[
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
])
data_cfg = dict(
image_size=512,
base_size=256,
base_sigma=2,
heatmap_size=[128],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
num_scales=1,
scale_aware_sigma=False,
)
# model settings
model = dict(
type='AssociativeEmbedding',
pretrained='https://download.openmmlab.com/mmpose/'
'pretrain_models/hrnet_w32-36af842e.pth',
backbone=dict(
type='HRNet',
in_channels=3,
extra=dict(
stage1=dict(
num_modules=1,
num_branches=1,
block='BOTTLENECK',
num_blocks=(4, ),
num_channels=(64, )),
stage2=dict(
num_modules=1,
num_branches=2,
block='BASIC',
num_blocks=(4, 4),
num_channels=(32, 64)),
stage3=dict(
num_modules=4,
num_branches=3,
block='BASIC',
num_blocks=(4, 4, 4),
num_channels=(32, 64, 128)),
stage4=dict(
num_modules=3,
num_branches=4,
block='BASIC',
num_blocks=(4, 4, 4, 4),
num_channels=(32, 64, 128, 256))),
),
keypoint_head=dict(
type='AESimpleHead',
in_channels=32,
num_joints=17,
num_deconv_layers=0,
tag_per_joint=True,
with_ae_loss=[True],
extra=dict(final_conv_kernel=1, ),
loss_keypoint=dict(
type='MultiLossFactory',
num_joints=17,
num_stages=1,
ae_loss_type='exp',
with_ae_loss=[True],
push_loss_factor=[0.001],
pull_loss_factor=[0.001],
with_heatmaps_loss=[True],
heatmaps_loss_factor=[1.0])),
train_cfg=dict(),
test_cfg=dict(
num_joints=channel_cfg['dataset_joints'],
max_num_people=30,
scale_factor=[1],
with_heatmaps=[True],
with_ae=[True],
project2image=True,
align_corners=False,
nms_kernel=5,
nms_padding=2,
tag_per_joint=True,
detection_threshold=0.1,
tag_threshold=1,
use_detection_val=True,
ignore_too_much=False,
adjust=True,
refine=True,
flip_test=True))
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='BottomUpRandomAffine',
rot_factor=30,
scale_factor=[0.75, 1.5],
scale_type='short',
trans_factor=40),
dict(type='BottomUpRandomFlip', flip_prob=0.5),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(
type='BottomUpGenerateTarget',
sigma=2,
max_num_people=30,
),
dict(
type='Collect',
keys=['img', 'joints', 'targets', 'masks'],
meta_keys=[]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='BottomUpGetImgSize', test_scale_factor=[1]),
dict(
type='BottomUpResizeAlign',
transforms=[
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
]),
dict(
type='Collect',
keys=['img'],
meta_keys=[
'image_file', 'aug_data', 'test_scale_factor', 'base_size',
'center', 'scale', 'flip_index'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
workers_per_gpu=2,
train_dataloader=dict(samples_per_gpu=24),
val_dataloader=dict(samples_per_gpu=1),
test_dataloader=dict(samples_per_gpu=1),
train=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
img_prefix=f'{data_root}/train2017/',
data_cfg=data_cfg,
pipeline=train_pipeline,
dataset_info={{_base_.dataset_info}}),
val=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=val_pipeline,
dataset_info={{_base_.dataset_info}}),
test=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=test_pipeline,
dataset_info={{_base_.dataset_info}}),
)
configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_w32_coco_512x512_udp.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
checkpoint_config = dict(interval=50)
evaluation = dict(interval=50, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=0.0015,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[200, 260])
total_epochs = 300
channel_cfg = dict(
dataset_joints=17,
dataset_channel=[
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
],
inference_channel=[
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
])
data_cfg = dict(
image_size=512,
base_size=256,
base_sigma=2,
heatmap_size=[128],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
num_scales=1,
scale_aware_sigma=False,
)
# model settings
model = dict(
type='AssociativeEmbedding',
pretrained='https://download.openmmlab.com/mmpose/'
'pretrain_models/hrnet_w32-36af842e.pth',
backbone=dict(
type='HRNet',
in_channels=3,
extra=dict(
stage1=dict(
num_modules=1,
num_branches=1,
block='BOTTLENECK',
num_blocks=(4, ),
num_channels=(64, )),
stage2=dict(
num_modules=1,
num_branches=2,
block='BASIC',
num_blocks=(4, 4),
num_channels=(32, 64)),
stage3=dict(
num_modules=4,
num_branches=3,
block='BASIC',
num_blocks=(4, 4, 4),
num_channels=(32, 64, 128)),
stage4=dict(
num_modules=3,
num_branches=4,
block='BASIC',
num_blocks=(4, 4, 4, 4),
num_channels=(32, 64, 128, 256))),
),
keypoint_head=dict(
type='AESimpleHead',
in_channels=32,
num_joints=17,
num_deconv_layers=0,
tag_per_joint=True,
with_ae_loss=[True],
extra=dict(final_conv_kernel=1, ),
loss_keypoint=dict(
type='MultiLossFactory',
num_joints=17,
num_stages=1,
ae_loss_type='exp',
with_ae_loss=[True],
push_loss_factor=[0.001],
pull_loss_factor=[0.001],
with_heatmaps_loss=[True],
heatmaps_loss_factor=[1.0])),
train_cfg=dict(),
test_cfg=dict(
num_joints=channel_cfg['dataset_joints'],
max_num_people=30,
scale_factor=[1],
with_heatmaps=[True],
with_ae=[True],
project2image=False,
align_corners=True,
nms_kernel=5,
nms_padding=2,
tag_per_joint=True,
detection_threshold=0.1,
tag_threshold=1,
use_detection_val=True,
ignore_too_much=False,
adjust=True,
refine=True,
flip_test=True,
use_udp=True))
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='BottomUpRandomAffine',
rot_factor=30,
scale_factor=[0.75, 1.5],
scale_type='short',
trans_factor=40,
use_udp=True),
dict(type='BottomUpRandomFlip', flip_prob=0.5),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(
type='BottomUpGenerateTarget',
sigma=2,
max_num_people=30,
use_udp=True,
),
dict(
type='Collect',
keys=['img', 'joints', 'targets', 'masks'],
meta_keys=[]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='BottomUpGetImgSize', test_scale_factor=[1], use_udp=True),
dict(
type='BottomUpResizeAlign',
transforms=[
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225])
],
use_udp=True),
dict(
type='Collect',
keys=['img'],
meta_keys=[
'image_file', 'aug_data', 'test_scale_factor', 'base_size',
'center', 'scale', 'flip_index'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
workers_per_gpu=2,
train_dataloader=dict(samples_per_gpu=24),
val_dataloader=dict(samples_per_gpu=1),
test_dataloader=dict(samples_per_gpu=1),
train=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
img_prefix=f'{data_root}/train2017/',
data_cfg=data_cfg,
pipeline=train_pipeline,
dataset_info={{_base_.dataset_info}}),
val=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=val_pipeline,
dataset_info={{_base_.dataset_info}}),
test=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=test_pipeline,
dataset_info={{_base_.dataset_info}}),
)
configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_w32_coco_640x640.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
checkpoint_config = dict(interval=50)
evaluation = dict(interval=50, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=0.0015,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[200, 260])
total_epochs = 300
channel_cfg = dict(
dataset_joints=17,
dataset_channel=[
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
],
inference_channel=[
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
])
data_cfg = dict(
image_size=640,
base_size=320,
base_sigma=2,
heatmap_size=[160],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
num_scales=1,
scale_aware_sigma=False,
)
# model settings
model = dict(
type='AssociativeEmbedding',
pretrained='https://download.openmmlab.com/mmpose/'
'pretrain_models/hrnet_w32-36af842e.pth',
backbone=dict(
type='HRNet',
in_channels=3,
extra=dict(
stage1=dict(
num_modules=1,
num_branches=1,
block='BOTTLENECK',
num_blocks=(4, ),
num_channels=(64, )),
stage2=dict(
num_modules=1,
num_branches=2,
block='BASIC',
num_blocks=(4, 4),
num_channels=(32, 64)),
stage3=dict(
num_modules=4,
num_branches=3,
block='BASIC',
num_blocks=(4, 4, 4),
num_channels=(32, 64, 128)),
stage4=dict(
num_modules=3,
num_branches=4,
block='BASIC',
num_blocks=(4, 4, 4, 4),
num_channels=(32, 64, 128, 256))),
),
keypoint_head=dict(
type='AESimpleHead',
in_channels=32,
num_joints=17,
num_deconv_layers=0,
tag_per_joint=True,
with_ae_loss=[True],
extra=dict(final_conv_kernel=1, ),
loss_keypoint=dict(
type='MultiLossFactory',
num_joints=17,
num_stages=1,
ae_loss_type='exp',
with_ae_loss=[True],
push_loss_factor=[0.001],
pull_loss_factor=[0.001],
with_heatmaps_loss=[True],
heatmaps_loss_factor=[1.0])),
train_cfg=dict(),
test_cfg=dict(
num_joints=channel_cfg['dataset_joints'],
max_num_people=30,
scale_factor=[1],
with_heatmaps=[True],
with_ae=[True],
project2image=True,
align_corners=False,
nms_kernel=5,
nms_padding=2,
tag_per_joint=True,
detection_threshold=0.1,
tag_threshold=1,
use_detection_val=True,
ignore_too_much=False,
adjust=True,
refine=True,
flip_test=True))
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='BottomUpRandomAffine',
rot_factor=30,
scale_factor=[0.75, 1.5],
scale_type='short',
trans_factor=40),
dict(type='BottomUpRandomFlip', flip_prob=0.5),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(
type='BottomUpGenerateTarget',
sigma=2,
max_num_people=30,
),
dict(
type='Collect',
keys=['img', 'joints', 'targets', 'masks'],
meta_keys=[]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='BottomUpGetImgSize', test_scale_factor=[1]),
dict(
type='BottomUpResizeAlign',
transforms=[
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
]),
dict(
type='Collect',
keys=['img'],
meta_keys=[
'image_file', 'aug_data', 'test_scale_factor', 'base_size',
'center', 'scale', 'flip_index'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
workers_per_gpu=2,
train_dataloader=dict(samples_per_gpu=16),
val_dataloader=dict(samples_per_gpu=1),
test_dataloader=dict(samples_per_gpu=1),
train=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
img_prefix=f'{data_root}/train2017/',
data_cfg=data_cfg,
pipeline=train_pipeline,
dataset_info={{_base_.dataset_info}}),
val=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=val_pipeline,
dataset_info={{_base_.dataset_info}}),
test=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=test_pipeline,
dataset_info={{_base_.dataset_info}}),
)
configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_w32_coco_640x640_udp.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
checkpoint_config = dict(interval=50)
evaluation = dict(interval=50, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=0.0015,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[200, 260])
total_epochs = 300
channel_cfg = dict(
dataset_joints=17,
dataset_channel=[
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
],
inference_channel=[
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
])
data_cfg = dict(
image_size=640,
base_size=320,
base_sigma=2,
heatmap_size=[160],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
num_scales=1,
scale_aware_sigma=False,
)
# model settings
model = dict(
type='AssociativeEmbedding',
pretrained='https://download.openmmlab.com/mmpose/'
'pretrain_models/hrnet_w32-36af842e.pth',
backbone=dict(
type='HRNet',
in_channels=3,
extra=dict(
stage1=dict(
num_modules=1,
num_branches=1,
block='BOTTLENECK',
num_blocks=(4, ),
num_channels=(64, )),
stage2=dict(
num_modules=1,
num_branches=2,
block='BASIC',
num_blocks=(4, 4),
num_channels=(32, 64)),
stage3=dict(
num_modules=4,
num_branches=3,
block='BASIC',
num_blocks=(4, 4, 4),
num_channels=(32, 64, 128)),
stage4=dict(
num_modules=3,
num_branches=4,
block='BASIC',
num_blocks=(4, 4, 4, 4),
num_channels=(32, 64, 128, 256))),
),
keypoint_head=dict(
type='AESimpleHead',
in_channels=32,
num_joints=17,
num_deconv_layers=0,
tag_per_joint=True,
with_ae_loss=[True],
extra=dict(final_conv_kernel=1, ),
loss_keypoint=dict(
type='MultiLossFactory',
num_joints=17,
num_stages=1,
ae_loss_type='exp',
with_ae_loss=[True],
push_loss_factor=[0.001],
pull_loss_factor=[0.001],
with_heatmaps_loss=[True],
heatmaps_loss_factor=[1.0])),
train_cfg=dict(),
test_cfg=dict(
num_joints=channel_cfg['dataset_joints'],
max_num_people=30,
scale_factor=[1],
with_heatmaps=[True],
with_ae=[True],
project2image=False,
align_corners=True,
nms_kernel=5,
nms_padding=2,
tag_per_joint=True,
detection_threshold=0.1,
tag_threshold=1,
use_detection_val=True,
ignore_too_much=False,
adjust=True,
refine=True,
flip_test=True,
use_udp=True))
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='BottomUpRandomAffine',
rot_factor=30,
scale_factor=[0.75, 1.5],
scale_type='short',
trans_factor=40,
use_udp=True),
dict(type='BottomUpRandomFlip', flip_prob=0.5),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(
type='BottomUpGenerateTarget',
sigma=2,
max_num_people=30,
use_udp=True,
),
dict(
type='Collect',
keys=['img', 'joints', 'targets', 'masks'],
meta_keys=[]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='BottomUpGetImgSize', test_scale_factor=[1], use_udp=True),
dict(
type='BottomUpResizeAlign',
transforms=[
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225])
],
use_udp=True),
dict(
type='Collect',
keys=['img'],
meta_keys=[
'image_file', 'aug_data', 'test_scale_factor', 'base_size',
'center', 'scale', 'flip_index'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
workers_per_gpu=2,
train_dataloader=dict(samples_per_gpu=16),
val_dataloader=dict(samples_per_gpu=1),
test_dataloader=dict(samples_per_gpu=1),
train=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
img_prefix=f'{data_root}/train2017/',
data_cfg=data_cfg,
pipeline=train_pipeline,
dataset_info={{_base_.dataset_info}}),
val=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=val_pipeline,
dataset_info={{_base_.dataset_info}}),
test=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=test_pipeline,
dataset_info={{_base_.dataset_info}}),
)
configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_w48_coco_512x512.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
checkpoint_config = dict(interval=50)
evaluation = dict(interval=50, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=0.0015,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[200, 260])
total_epochs = 300
channel_cfg = dict(
dataset_joints=17,
dataset_channel=[
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
],
inference_channel=[
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
])
data_cfg = dict(
image_size=512,
base_size=256,
base_sigma=2,
heatmap_size=[128],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
num_scales=1,
scale_aware_sigma=False,
)
# model settings
model = dict(
type='AssociativeEmbedding',
pretrained='https://download.openmmlab.com/mmpose/'
'pretrain_models/hrnet_w48-8ef0771d.pth',
backbone=dict(
type='HRNet',
in_channels=3,
extra=dict(
stage1=dict(
num_modules=1,
num_branches=1,
block='BOTTLENECK',
num_blocks=(4, ),
num_channels=(64, )),
stage2=dict(
num_modules=1,
num_branches=2,
block='BASIC',
num_blocks=(4, 4),
num_channels=(48, 96)),
stage3=dict(
num_modules=4,
num_branches=3,
block='BASIC',
num_blocks=(4, 4, 4),
num_channels=(48, 96, 192)),
stage4=dict(
num_modules=3,
num_branches=4,
block='BASIC',
num_blocks=(4, 4, 4, 4),
num_channels=(48, 96, 192, 384))),
),
keypoint_head=dict(
type='AESimpleHead',
in_channels=48,
num_joints=17,
num_deconv_layers=0,
tag_per_joint=True,
with_ae_loss=[True],
extra=dict(final_conv_kernel=1, ),
loss_keypoint=dict(
type='MultiLossFactory',
num_joints=17,
num_stages=1,
ae_loss_type='exp',
with_ae_loss=[True],
push_loss_factor=[0.001],
pull_loss_factor=[0.001],
with_heatmaps_loss=[True],
heatmaps_loss_factor=[1.0])),
train_cfg=dict(),
test_cfg=dict(
num_joints=channel_cfg['dataset_joints'],
max_num_people=30,
scale_factor=[1],
with_heatmaps=[True],
with_ae=[True],
project2image=True,
align_corners=False,
nms_kernel=5,
nms_padding=2,
tag_per_joint=True,
detection_threshold=0.1,
tag_threshold=1,
use_detection_val=True,
ignore_too_much=False,
adjust=True,
refine=True,
flip_test=True))
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='BottomUpRandomAffine',
rot_factor=30,
scale_factor=[0.75, 1.5],
scale_type='short',
trans_factor=40),
dict(type='BottomUpRandomFlip', flip_prob=0.5),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(
type='BottomUpGenerateTarget',
sigma=2,
max_num_people=30,
),
dict(
type='Collect',
keys=['img', 'joints', 'targets', 'masks'],
meta_keys=[]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='BottomUpGetImgSize', test_scale_factor=[1]),
dict(
type='BottomUpResizeAlign',
transforms=[
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
]),
dict(
type='Collect',
keys=['img'],
meta_keys=[
'image_file', 'aug_data', 'test_scale_factor', 'base_size',
'center', 'scale', 'flip_index'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
workers_per_gpu=2,
train_dataloader=dict(samples_per_gpu=16),
val_dataloader=dict(samples_per_gpu=1),
test_dataloader=dict(samples_per_gpu=1),
train=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
img_prefix=f'{data_root}/train2017/',
data_cfg=data_cfg,
pipeline=train_pipeline,
dataset_info={{_base_.dataset_info}}),
val=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=val_pipeline,
dataset_info={{_base_.dataset_info}}),
test=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=test_pipeline,
dataset_info={{_base_.dataset_info}}),
)
configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_w48_coco_512x512_udp.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
checkpoint_config = dict(interval=50)
evaluation = dict(interval=50, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=0.0015,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[200, 260])
total_epochs = 300
channel_cfg = dict(
dataset_joints=17,
dataset_channel=[
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
],
inference_channel=[
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
])
data_cfg = dict(
image_size=512,
base_size=256,
base_sigma=2,
heatmap_size=[128],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
num_scales=1,
scale_aware_sigma=False,
)
# model settings
model = dict(
type='AssociativeEmbedding',
pretrained='https://download.openmmlab.com/mmpose/'
'pretrain_models/hrnet_w48-8ef0771d.pth',
backbone=dict(
type='HRNet',
in_channels=3,
extra=dict(
stage1=dict(
num_modules=1,
num_branches=1,
block='BOTTLENECK',
num_blocks=(4, ),
num_channels=(64, )),
stage2=dict(
num_modules=1,
num_branches=2,
block='BASIC',
num_blocks=(4, 4),
num_channels=(48, 96)),
stage3=dict(
num_modules=4,
num_branches=3,
block='BASIC',
num_blocks=(4, 4, 4),
num_channels=(48, 96, 192)),
stage4=dict(
num_modules=3,
num_branches=4,
block='BASIC',
num_blocks=(4, 4, 4, 4),
num_channels=(48, 96, 192, 384))),
),
keypoint_head=dict(
type='AESimpleHead',
in_channels=48,
num_joints=17,
num_deconv_layers=0,
tag_per_joint=True,
with_ae_loss=[True],
extra=dict(final_conv_kernel=1, ),
loss_keypoint=dict(
type='MultiLossFactory',
num_joints=17,
num_stages=1,
ae_loss_type='exp',
with_ae_loss=[True],
push_loss_factor=[0.001],
pull_loss_factor=[0.001],
with_heatmaps_loss=[True],
heatmaps_loss_factor=[1.0])),
train_cfg=dict(),
test_cfg=dict(
num_joints=channel_cfg['dataset_joints'],
max_num_people=30,
scale_factor=[1],
with_heatmaps=[True],
with_ae=[True],
project2image=False,
align_corners=True,
nms_kernel=5,
nms_padding=2,
tag_per_joint=True,
detection_threshold=0.1,
tag_threshold=1,
use_detection_val=True,
ignore_too_much=False,
adjust=True,
refine=True,
flip_test=True,
use_udp=True))
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='BottomUpRandomAffine',
rot_factor=30,
scale_factor=[0.75, 1.5],
scale_type='short',
trans_factor=40,
use_udp=True),
dict(type='BottomUpRandomFlip', flip_prob=0.5),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(
type='BottomUpGenerateTarget',
sigma=2,
max_num_people=30,
use_udp=True,
),
dict(
type='Collect',
keys=['img', 'joints', 'targets', 'masks'],
meta_keys=[]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='BottomUpGetImgSize', test_scale_factor=[1], use_udp=True),
dict(
type='BottomUpResizeAlign',
transforms=[
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225])
],
use_udp=True),
dict(
type='Collect',
keys=['img'],
meta_keys=[
'image_file', 'aug_data', 'test_scale_factor', 'base_size',
'center', 'scale', 'flip_index'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
workers_per_gpu=2,
train_dataloader=dict(samples_per_gpu=16),
val_dataloader=dict(samples_per_gpu=1),
test_dataloader=dict(samples_per_gpu=1),
train=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
img_prefix=f'{data_root}/train2017/',
data_cfg=data_cfg,
pipeline=train_pipeline,
dataset_info={{_base_.dataset_info}}),
val=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=val_pipeline,
dataset_info={{_base_.dataset_info}}),
test=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=test_pipeline,
dataset_info={{_base_.dataset_info}}),
)
configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_w48_coco_640x640.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
checkpoint_config = dict(interval=50)
evaluation = dict(interval=50, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=0.0015,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[200, 260])
total_epochs = 300
channel_cfg = dict(
dataset_joints=17,
dataset_channel=[
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
],
inference_channel=[
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
])
data_cfg = dict(
image_size=640,
base_size=320,
base_sigma=2,
heatmap_size=[160],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
num_scales=1,
scale_aware_sigma=False,
)
# model settings
model = dict(
type='AssociativeEmbedding',
pretrained='https://download.openmmlab.com/mmpose/'
'pretrain_models/hrnet_w48-8ef0771d.pth',
backbone=dict(
type='HRNet',
in_channels=3,
extra=dict(
stage1=dict(
num_modules=1,
num_branches=1,
block='BOTTLENECK',
num_blocks=(4, ),
num_channels=(64, )),
stage2=dict(
num_modules=1,
num_branches=2,
block='BASIC',
num_blocks=(4, 4),
num_channels=(48, 96)),
stage3=dict(
num_modules=4,
num_branches=3,
block='BASIC',
num_blocks=(4, 4, 4),
num_channels=(48, 96, 192)),
stage4=dict(
num_modules=3,
num_branches=4,
block='BASIC',
num_blocks=(4, 4, 4, 4),
num_channels=(48, 96, 192, 384))),
),
keypoint_head=dict(
type='AESimpleHead',
in_channels=48,
num_joints=17,
num_deconv_layers=0,
tag_per_joint=True,
with_ae_loss=[True],
extra=dict(final_conv_kernel=1, ),
loss_keypoint=dict(
type='MultiLossFactory',
num_joints=17,
num_stages=1,
ae_loss_type='exp',
with_ae_loss=[True],
push_loss_factor=[0.001],
pull_loss_factor=[0.001],
with_heatmaps_loss=[True],
heatmaps_loss_factor=[1.0])),
train_cfg=dict(),
test_cfg=dict(
num_joints=channel_cfg['dataset_joints'],
max_num_people=30,
scale_factor=[1],
with_heatmaps=[True],
with_ae=[True],
project2image=True,
align_corners=False,
nms_kernel=5,
nms_padding=2,
tag_per_joint=True,
detection_threshold=0.1,
tag_threshold=1,
use_detection_val=True,
ignore_too_much=False,
adjust=True,
refine=True,
flip_test=True))
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='BottomUpRandomAffine',
rot_factor=30,
scale_factor=[0.75, 1.5],
scale_type='short',
trans_factor=40),
dict(type='BottomUpRandomFlip', flip_prob=0.5),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(
type='BottomUpGenerateTarget',
sigma=2,
max_num_people=30,
),
dict(
type='Collect',
keys=['img', 'joints', 'targets', 'masks'],
meta_keys=[]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='BottomUpGetImgSize', test_scale_factor=[1]),
dict(
type='BottomUpResizeAlign',
transforms=[
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
]),
dict(
type='Collect',
keys=['img'],
meta_keys=[
'image_file', 'aug_data', 'test_scale_factor', 'base_size',
'center', 'scale', 'flip_index'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
workers_per_gpu=2,
train_dataloader=dict(samples_per_gpu=8),
val_dataloader=dict(samples_per_gpu=1),
test_dataloader=dict(samples_per_gpu=1),
train=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
img_prefix=f'{data_root}/train2017/',
data_cfg=data_cfg,
pipeline=train_pipeline,
dataset_info={{_base_.dataset_info}}),
val=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=val_pipeline,
dataset_info={{_base_.dataset_info}}),
test=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=test_pipeline,
dataset_info={{_base_.dataset_info}}),
)
configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/hrnet_w48_coco_640x640_udp.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
checkpoint_config = dict(interval=50)
evaluation = dict(interval=50, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=0.0015,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[200, 260])
total_epochs = 300
channel_cfg = dict(
dataset_joints=17,
dataset_channel=[
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
],
inference_channel=[
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
])
data_cfg = dict(
image_size=640,
base_size=320,
base_sigma=2,
heatmap_size=[160],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
num_scales=1,
scale_aware_sigma=False,
)
# model settings
model = dict(
type='AssociativeEmbedding',
pretrained='https://download.openmmlab.com/mmpose/'
'pretrain_models/hrnet_w48-8ef0771d.pth',
backbone=dict(
type='HRNet',
in_channels=3,
extra=dict(
stage1=dict(
num_modules=1,
num_branches=1,
block='BOTTLENECK',
num_blocks=(4, ),
num_channels=(64, )),
stage2=dict(
num_modules=1,
num_branches=2,
block='BASIC',
num_blocks=(4, 4),
num_channels=(48, 96)),
stage3=dict(
num_modules=4,
num_branches=3,
block='BASIC',
num_blocks=(4, 4, 4),
num_channels=(48, 96, 192)),
stage4=dict(
num_modules=3,
num_branches=4,
block='BASIC',
num_blocks=(4, 4, 4, 4),
num_channels=(48, 96, 192, 384))),
),
keypoint_head=dict(
type='AESimpleHead',
in_channels=48,
num_joints=17,
num_deconv_layers=0,
tag_per_joint=True,
with_ae_loss=[True],
extra=dict(final_conv_kernel=1, ),
loss_keypoint=dict(
type='MultiLossFactory',
num_joints=17,
num_stages=1,
ae_loss_type='exp',
with_ae_loss=[True],
push_loss_factor=[0.001],
pull_loss_factor=[0.001],
with_heatmaps_loss=[True],
heatmaps_loss_factor=[1.0])),
train_cfg=dict(),
test_cfg=dict(
num_joints=channel_cfg['dataset_joints'],
max_num_people=30,
scale_factor=[1],
with_heatmaps=[True],
with_ae=[True],
project2image=False,
align_corners=True,
nms_kernel=5,
nms_padding=2,
tag_per_joint=True,
detection_threshold=0.1,
tag_threshold=1,
use_detection_val=True,
ignore_too_much=False,
adjust=True,
refine=True,
flip_test=True,
use_udp=True))
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='BottomUpRandomAffine',
rot_factor=30,
scale_factor=[0.75, 1.5],
scale_type='short',
trans_factor=40,
use_udp=True),
dict(type='BottomUpRandomFlip', flip_prob=0.5),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(
type='BottomUpGenerateTarget',
sigma=2,
max_num_people=30,
use_udp=True,
),
dict(
type='Collect',
keys=['img', 'joints', 'targets', 'masks'],
meta_keys=[]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='BottomUpGetImgSize', test_scale_factor=[1], use_udp=True),
dict(
type='BottomUpResizeAlign',
transforms=[
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225])
],
use_udp=True),
dict(
type='Collect',
keys=['img'],
meta_keys=[
'image_file', 'aug_data', 'test_scale_factor', 'base_size',
'center', 'scale', 'flip_index'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
workers_per_gpu=2,
train_dataloader=dict(samples_per_gpu=8),
val_dataloader=dict(samples_per_gpu=1),
test_dataloader=dict(samples_per_gpu=1),
train=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
img_prefix=f'{data_root}/train2017/',
data_cfg=data_cfg,
pipeline=train_pipeline,
dataset_info={{_base_.dataset_info}}),
val=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=val_pipeline,
dataset_info={{_base_.dataset_info}}),
test=dict(
type='BottomUpCocoDataset',
ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
img_prefix=f'{data_root}/val2017/',
data_cfg=data_cfg,
pipeline=test_pipeline,
dataset_info={{_base_.dataset_info}}),
)
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment