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 2288 additions and 0 deletions
configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnest101_coco_384x288.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
evaluation = dict(interval=10, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=5e-4,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[170, 200])
total_epochs = 210
channel_cfg = dict(
num_output_channels=17,
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
])
# model settings
model = dict(
type='TopDown',
pretrained='mmcls://resnest101',
backbone=dict(type='ResNeSt', depth=101),
keypoint_head=dict(
type='TopdownHeatmapSimpleHead',
in_channels=2048,
out_channels=channel_cfg['num_output_channels'],
loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
train_cfg=dict(),
test_cfg=dict(
flip_test=True,
post_process='default',
shift_heatmap=True,
modulate_kernel=11))
data_cfg = dict(
image_size=[288, 384],
heatmap_size=[72, 96],
num_output_channels=channel_cfg['num_output_channels'],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
soft_nms=False,
nms_thr=1.0,
oks_thr=0.9,
vis_thr=0.2,
use_gt_bbox=False,
det_bbox_thr=0.0,
bbox_file='data/coco/person_detection_results/'
'COCO_val2017_detections_AP_H_56_person.json',
)
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownRandomShiftBboxCenter', shift_factor=0.16, prob=0.3),
dict(type='TopDownRandomFlip', flip_prob=0.5),
dict(
type='TopDownHalfBodyTransform',
num_joints_half_body=8,
prob_half_body=0.3),
dict(
type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
dict(type='TopDownAffine'),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(type='TopDownGenerateTarget', sigma=3),
dict(
type='Collect',
keys=['img', 'target', 'target_weight'],
meta_keys=[
'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
'rotation', 'bbox_score', 'flip_pairs'
]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownAffine'),
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', 'center', 'scale', 'rotation', 'bbox_score',
'flip_pairs'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
samples_per_gpu=32,
workers_per_gpu=2,
val_dataloader=dict(samples_per_gpu=32),
test_dataloader=dict(samples_per_gpu=32),
train=dict(
type='TopDownCocoDataset',
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='TopDownCocoDataset',
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='TopDownCocoDataset',
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/topdown_heatmap/coco/resnest200_coco_256x192.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
evaluation = dict(interval=10, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=5e-4,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[170, 200])
total_epochs = 210
channel_cfg = dict(
num_output_channels=17,
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
])
# model settings
model = dict(
type='TopDown',
pretrained='mmcls://resnest200',
backbone=dict(type='ResNeSt', depth=200),
keypoint_head=dict(
type='TopdownHeatmapSimpleHead',
in_channels=2048,
out_channels=channel_cfg['num_output_channels'],
loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
train_cfg=dict(),
test_cfg=dict(
flip_test=True,
post_process='default',
shift_heatmap=True,
modulate_kernel=11))
data_cfg = dict(
image_size=[192, 256],
heatmap_size=[48, 64],
num_output_channels=channel_cfg['num_output_channels'],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
soft_nms=False,
nms_thr=1.0,
oks_thr=0.9,
vis_thr=0.2,
use_gt_bbox=False,
det_bbox_thr=0.0,
bbox_file='data/coco/person_detection_results/'
'COCO_val2017_detections_AP_H_56_person.json',
)
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownRandomShiftBboxCenter', shift_factor=0.16, prob=0.3),
dict(type='TopDownRandomFlip', flip_prob=0.5),
dict(
type='TopDownHalfBodyTransform',
num_joints_half_body=8,
prob_half_body=0.3),
dict(
type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
dict(type='TopDownAffine'),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(type='TopDownGenerateTarget', sigma=2),
dict(
type='Collect',
keys=['img', 'target', 'target_weight'],
meta_keys=[
'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
'rotation', 'bbox_score', 'flip_pairs'
]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownAffine'),
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', 'center', 'scale', 'rotation', 'bbox_score',
'flip_pairs'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
samples_per_gpu=64,
workers_per_gpu=2,
val_dataloader=dict(samples_per_gpu=32),
test_dataloader=dict(samples_per_gpu=32),
train=dict(
type='TopDownCocoDataset',
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='TopDownCocoDataset',
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='TopDownCocoDataset',
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/topdown_heatmap/coco/resnest200_coco_384x288.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
evaluation = dict(interval=10, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=5e-4,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[170, 200])
total_epochs = 210
channel_cfg = dict(
num_output_channels=17,
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
])
# model settings
model = dict(
type='TopDown',
pretrained='mmcls://resnest200',
backbone=dict(type='ResNeSt', depth=200),
keypoint_head=dict(
type='TopdownHeatmapSimpleHead',
in_channels=2048,
out_channels=channel_cfg['num_output_channels'],
loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
train_cfg=dict(),
test_cfg=dict(
flip_test=True,
post_process='default',
shift_heatmap=True,
modulate_kernel=11))
data_cfg = dict(
image_size=[288, 384],
heatmap_size=[72, 96],
num_output_channels=channel_cfg['num_output_channels'],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
soft_nms=False,
nms_thr=1.0,
oks_thr=0.9,
vis_thr=0.2,
use_gt_bbox=False,
det_bbox_thr=0.0,
bbox_file='data/coco/person_detection_results/'
'COCO_val2017_detections_AP_H_56_person.json',
)
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownRandomShiftBboxCenter', shift_factor=0.16, prob=0.3),
dict(type='TopDownRandomFlip', flip_prob=0.5),
dict(
type='TopDownHalfBodyTransform',
num_joints_half_body=8,
prob_half_body=0.3),
dict(
type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
dict(type='TopDownAffine'),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(type='TopDownGenerateTarget', sigma=3),
dict(
type='Collect',
keys=['img', 'target', 'target_weight'],
meta_keys=[
'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
'rotation', 'bbox_score', 'flip_pairs'
]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownAffine'),
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', 'center', 'scale', 'rotation', 'bbox_score',
'flip_pairs'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
samples_per_gpu=16,
workers_per_gpu=2,
val_dataloader=dict(samples_per_gpu=16),
test_dataloader=dict(samples_per_gpu=16),
train=dict(
type='TopDownCocoDataset',
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='TopDownCocoDataset',
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='TopDownCocoDataset',
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/topdown_heatmap/coco/resnest269_coco_256x192.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
evaluation = dict(interval=10, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=5e-4,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[170, 200])
total_epochs = 210
channel_cfg = dict(
num_output_channels=17,
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
])
# model settings
model = dict(
type='TopDown',
pretrained='mmcls://resnest269',
backbone=dict(type='ResNeSt', depth=269),
keypoint_head=dict(
type='TopdownHeatmapSimpleHead',
in_channels=2048,
out_channels=channel_cfg['num_output_channels'],
loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
train_cfg=dict(),
test_cfg=dict(
flip_test=True,
post_process='default',
shift_heatmap=True,
modulate_kernel=11))
data_cfg = dict(
image_size=[192, 256],
heatmap_size=[48, 64],
num_output_channels=channel_cfg['num_output_channels'],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
soft_nms=False,
nms_thr=1.0,
oks_thr=0.9,
vis_thr=0.2,
use_gt_bbox=False,
det_bbox_thr=0.0,
bbox_file='data/coco/person_detection_results/'
'COCO_val2017_detections_AP_H_56_person.json',
)
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownRandomShiftBboxCenter', shift_factor=0.16, prob=0.3),
dict(type='TopDownRandomFlip', flip_prob=0.5),
dict(
type='TopDownHalfBodyTransform',
num_joints_half_body=8,
prob_half_body=0.3),
dict(
type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
dict(type='TopDownAffine'),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(type='TopDownGenerateTarget', sigma=2),
dict(
type='Collect',
keys=['img', 'target', 'target_weight'],
meta_keys=[
'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
'rotation', 'bbox_score', 'flip_pairs'
]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownAffine'),
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', 'center', 'scale', 'rotation', 'bbox_score',
'flip_pairs'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
samples_per_gpu=32,
workers_per_gpu=2,
val_dataloader=dict(samples_per_gpu=32),
test_dataloader=dict(samples_per_gpu=32),
train=dict(
type='TopDownCocoDataset',
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='TopDownCocoDataset',
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='TopDownCocoDataset',
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/topdown_heatmap/coco/resnest269_coco_384x288.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
evaluation = dict(interval=10, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=5e-4,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[170, 200])
total_epochs = 210
channel_cfg = dict(
num_output_channels=17,
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
])
# model settings
model = dict(
type='TopDown',
pretrained='mmcls://resnest269',
backbone=dict(type='ResNeSt', depth=269),
keypoint_head=dict(
type='TopdownHeatmapSimpleHead',
in_channels=2048,
out_channels=channel_cfg['num_output_channels'],
loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
train_cfg=dict(),
test_cfg=dict(
flip_test=True,
post_process='default',
shift_heatmap=True,
modulate_kernel=11))
data_cfg = dict(
image_size=[288, 384],
heatmap_size=[72, 96],
num_output_channels=channel_cfg['num_output_channels'],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
soft_nms=False,
nms_thr=1.0,
oks_thr=0.9,
vis_thr=0.2,
use_gt_bbox=False,
det_bbox_thr=0.0,
bbox_file='data/coco/person_detection_results/'
'COCO_val2017_detections_AP_H_56_person.json',
)
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownRandomShiftBboxCenter', shift_factor=0.16, prob=0.3),
dict(type='TopDownRandomFlip', flip_prob=0.5),
dict(
type='TopDownHalfBodyTransform',
num_joints_half_body=8,
prob_half_body=0.3),
dict(
type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
dict(type='TopDownAffine'),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(type='TopDownGenerateTarget', sigma=3),
dict(
type='Collect',
keys=['img', 'target', 'target_weight'],
meta_keys=[
'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
'rotation', 'bbox_score', 'flip_pairs'
]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownAffine'),
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', 'center', 'scale', 'rotation', 'bbox_score',
'flip_pairs'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
samples_per_gpu=16,
workers_per_gpu=2,
val_dataloader=dict(samples_per_gpu=16),
test_dataloader=dict(samples_per_gpu=16),
train=dict(
type='TopDownCocoDataset',
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='TopDownCocoDataset',
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='TopDownCocoDataset',
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/topdown_heatmap/coco/resnest50_coco_256x192.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
evaluation = dict(interval=10, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=5e-4,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[170, 200])
total_epochs = 210
channel_cfg = dict(
num_output_channels=17,
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
])
# model settings
model = dict(
type='TopDown',
pretrained='mmcls://resnest50',
backbone=dict(type='ResNeSt', depth=50),
keypoint_head=dict(
type='TopdownHeatmapSimpleHead',
in_channels=2048,
out_channels=channel_cfg['num_output_channels'],
loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
train_cfg=dict(),
test_cfg=dict(
flip_test=True,
post_process='default',
shift_heatmap=True,
modulate_kernel=11))
data_cfg = dict(
image_size=[192, 256],
heatmap_size=[48, 64],
num_output_channels=channel_cfg['num_output_channels'],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
soft_nms=False,
nms_thr=1.0,
oks_thr=0.9,
vis_thr=0.2,
use_gt_bbox=False,
det_bbox_thr=0.0,
bbox_file='data/coco/person_detection_results/'
'COCO_val2017_detections_AP_H_56_person.json',
)
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownRandomShiftBboxCenter', shift_factor=0.16, prob=0.3),
dict(type='TopDownRandomFlip', flip_prob=0.5),
dict(
type='TopDownHalfBodyTransform',
num_joints_half_body=8,
prob_half_body=0.3),
dict(
type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
dict(type='TopDownAffine'),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(type='TopDownGenerateTarget', sigma=2),
dict(
type='Collect',
keys=['img', 'target', 'target_weight'],
meta_keys=[
'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
'rotation', 'bbox_score', 'flip_pairs'
]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownAffine'),
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', 'center', 'scale', 'rotation', 'bbox_score',
'flip_pairs'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
samples_per_gpu=64,
workers_per_gpu=2,
val_dataloader=dict(samples_per_gpu=32),
test_dataloader=dict(samples_per_gpu=32),
train=dict(
type='TopDownCocoDataset',
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='TopDownCocoDataset',
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='TopDownCocoDataset',
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/topdown_heatmap/coco/resnest50_coco_384x288.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
evaluation = dict(interval=10, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=5e-4,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[170, 200])
total_epochs = 210
channel_cfg = dict(
num_output_channels=17,
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
])
# model settings
model = dict(
type='TopDown',
pretrained='mmcls://resnest50',
backbone=dict(type='ResNeSt', depth=50),
keypoint_head=dict(
type='TopdownHeatmapSimpleHead',
in_channels=2048,
out_channels=channel_cfg['num_output_channels'],
loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
train_cfg=dict(),
test_cfg=dict(
flip_test=True,
post_process='default',
shift_heatmap=True,
modulate_kernel=11))
data_cfg = dict(
image_size=[288, 384],
heatmap_size=[72, 96],
num_output_channels=channel_cfg['num_output_channels'],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
soft_nms=False,
nms_thr=1.0,
oks_thr=0.9,
vis_thr=0.2,
use_gt_bbox=False,
det_bbox_thr=0.0,
bbox_file='data/coco/person_detection_results/'
'COCO_val2017_detections_AP_H_56_person.json',
)
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownRandomShiftBboxCenter', shift_factor=0.16, prob=0.3),
dict(type='TopDownRandomFlip', flip_prob=0.5),
dict(
type='TopDownHalfBodyTransform',
num_joints_half_body=8,
prob_half_body=0.3),
dict(
type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
dict(type='TopDownAffine'),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(type='TopDownGenerateTarget', sigma=3),
dict(
type='Collect',
keys=['img', 'target', 'target_weight'],
meta_keys=[
'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
'rotation', 'bbox_score', 'flip_pairs'
]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownAffine'),
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', 'center', 'scale', 'rotation', 'bbox_score',
'flip_pairs'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
samples_per_gpu=64,
workers_per_gpu=2,
val_dataloader=dict(samples_per_gpu=32),
test_dataloader=dict(samples_per_gpu=32),
train=dict(
type='TopDownCocoDataset',
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='TopDownCocoDataset',
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='TopDownCocoDataset',
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/topdown_heatmap/coco/resnest_coco.md 0 → 100755
View file @ 37c8cebc
<!-- [BACKBONE] -->
<details>
<summary align="right"><a href="https://arxiv.org/abs/2004.08955">ResNeSt (ArXiv'2020)</a></summary>
```bibtex
@article{zhang2020resnest,
title={ResNeSt: Split-Attention Networks},
author={Zhang, Hang and Wu, Chongruo and Zhang, Zhongyue and Zhu, Yi and Zhang, Zhi and Lin, Haibin and Sun, Yue and He, Tong and Muller, Jonas and Manmatha, R. and Li, Mu and Smola, Alexander},
journal={arXiv preprint arXiv:2004.08955},
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 with detector having human AP of 56.4 on COCO val2017 dataset
| Arch | Input Size | AP | AP<sup>50</sup> | AP<sup>75</sup> | AR | AR<sup>50</sup> | ckpt | log |
| :-------------------------------------------- | :--------: | :---: | :-------------: | :-------------: | :---: | :-------------: | :-------------------------------------------: | :-------------------------------------------: |
| [pose_resnest_50](/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnest50_coco_256x192.py) | 256x192 | 0.721 | 0.899 | 0.802 | 0.776 | 0.938 | [ckpt](https://download.openmmlab.com/mmpose/top_down/resnest/resnest50_coco_256x192-6e65eece_20210320.pth) | [log](https://download.openmmlab.com/mmpose/top_down/resnest/resnest50_coco_256x192_20210320.log.json) |
| [pose_resnest_50](/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnest50_coco_384x288.py) | 384x288 | 0.737 | 0.900 | 0.811 | 0.789 | 0.938 | [ckpt](https://download.openmmlab.com/mmpose/top_down/resnest/resnest50_coco_384x288-dcd20436_20210320.pth) | [log](https://download.openmmlab.com/mmpose/top_down/resnest/resnest50_coco_384x288_20210320.log.json) |
| [pose_resnest_101](/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnest101_coco_256x192.py) | 256x192 | 0.725 | 0.899 | 0.807 | 0.781 | 0.939 | [ckpt](https://download.openmmlab.com/mmpose/top_down/resnest/resnest101_coco_256x192-2ffcdc9d_20210320.pth) | [log](https://download.openmmlab.com/mmpose/top_down/resnest/resnest101_coco_256x192_20210320.log.json) |
| [pose_resnest_101](/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnest101_coco_384x288.py) | 384x288 | 0.746 | 0.906 | 0.820 | 0.798 | 0.943 | [ckpt](https://download.openmmlab.com/mmpose/top_down/resnest/resnest101_coco_384x288-80660658_20210320.pth) | [log](https://download.openmmlab.com/mmpose/top_down/resnest/resnest101_coco_384x288_20210320.log.json) |
| [pose_resnest_200](/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnest200_coco_256x192.py) | 256x192 | 0.732 | 0.905 | 0.812 | 0.787 | 0.942 | [ckpt](https://download.openmmlab.com/mmpose/top_down/resnest/resnest200_coco_256x192-db007a48_20210517.pth) | [log](https://download.openmmlab.com/mmpose/top_down/resnest/resnest200_coco_256x192_20210517.log.json) |
| [pose_resnest_200](/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnest200_coco_384x288.py) | 384x288 | 0.754 | 0.908 | 0.827 | 0.807 | 0.945 | [ckpt](https://download.openmmlab.com/mmpose/top_down/resnest/resnest200_coco_384x288-b5bb76cb_20210517.pth) | [log](https://download.openmmlab.com/mmpose/top_down/resnest/resnest200_coco_384x288_20210517.log.json) |
| [pose_resnest_269](/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnest269_coco_256x192.py) | 256x192 | 0.738 | 0.907 | 0.819 | 0.793 | 0.945 | [ckpt](https://download.openmmlab.com/mmpose/top_down/resnest/resnest269_coco_256x192-2a7882ac_20210517.pth) | [log](https://download.openmmlab.com/mmpose/top_down/resnest/resnest269_coco_256x192_20210517.log.json) |
| [pose_resnest_269](/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnest269_coco_384x288.py) | 384x288 | 0.755 | 0.908 | 0.828 | 0.806 | 0.943 | [ckpt](https://download.openmmlab.com/mmpose/top_down/resnest/resnest269_coco_384x288-b142b9fb_20210517.pth) | [log](https://download.openmmlab.com/mmpose/top_down/resnest/resnest269_coco_384x288_20210517.log.json) |
configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnest_coco.yml 0 → 100755
View file @ 37c8cebc
Collections:
- Name: ResNeSt
Paper:
Title: 'ResNeSt: Split-Attention Networks'
URL: https://arxiv.org/abs/2004.08955
README: https://github.com/open-mmlab/mmpose/blob/master/docs/en/papers/backbones/resnest.md
Models:
- Config: configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnest50_coco_256x192.py
In Collection: ResNeSt
Metadata:
Architecture: &id001
- ResNeSt
Training Data: COCO
Name: topdown_heatmap_resnest50_coco_256x192
Results:
- Dataset: COCO
Metrics:
AP: 0.721
AP@0.5: 0.899
AP@0.75: 0.802
AR: 0.776
AR@0.5: 0.938
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/top_down/resnest/resnest50_coco_256x192-6e65eece_20210320.pth
- Config: configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnest50_coco_384x288.py
In Collection: ResNeSt
Metadata:
Architecture: *id001
Training Data: COCO
Name: topdown_heatmap_resnest50_coco_384x288
Results:
- Dataset: COCO
Metrics:
AP: 0.737
AP@0.5: 0.9
AP@0.75: 0.811
AR: 0.789
AR@0.5: 0.938
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/top_down/resnest/resnest50_coco_384x288-dcd20436_20210320.pth
- Config: configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnest101_coco_256x192.py
In Collection: ResNeSt
Metadata:
Architecture: *id001
Training Data: COCO
Name: topdown_heatmap_resnest101_coco_256x192
Results:
- Dataset: COCO
Metrics:
AP: 0.725
AP@0.5: 0.899
AP@0.75: 0.807
AR: 0.781
AR@0.5: 0.939
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/top_down/resnest/resnest101_coco_256x192-2ffcdc9d_20210320.pth
- Config: configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnest101_coco_384x288.py
In Collection: ResNeSt
Metadata:
Architecture: *id001
Training Data: COCO
Name: topdown_heatmap_resnest101_coco_384x288
Results:
- Dataset: COCO
Metrics:
AP: 0.746
AP@0.5: 0.906
AP@0.75: 0.82
AR: 0.798
AR@0.5: 0.943
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/top_down/resnest/resnest101_coco_384x288-80660658_20210320.pth
- Config: configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnest200_coco_256x192.py
In Collection: ResNeSt
Metadata:
Architecture: *id001
Training Data: COCO
Name: topdown_heatmap_resnest200_coco_256x192
Results:
- Dataset: COCO
Metrics:
AP: 0.732
AP@0.5: 0.905
AP@0.75: 0.812
AR: 0.787
AR@0.5: 0.942
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/top_down/resnest/resnest200_coco_256x192-db007a48_20210517.pth
- Config: configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnest200_coco_384x288.py
In Collection: ResNeSt
Metadata:
Architecture: *id001
Training Data: COCO
Name: topdown_heatmap_resnest200_coco_384x288
Results:
- Dataset: COCO
Metrics:
AP: 0.754
AP@0.5: 0.908
AP@0.75: 0.827
AR: 0.807
AR@0.5: 0.945
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/top_down/resnest/resnest200_coco_384x288-b5bb76cb_20210517.pth
- Config: configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnest269_coco_256x192.py
In Collection: ResNeSt
Metadata:
Architecture: *id001
Training Data: COCO
Name: topdown_heatmap_resnest269_coco_256x192
Results:
- Dataset: COCO
Metrics:
AP: 0.738
AP@0.5: 0.907
AP@0.75: 0.819
AR: 0.793
AR@0.5: 0.945
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/top_down/resnest/resnest269_coco_256x192-2a7882ac_20210517.pth
- Config: configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnest269_coco_384x288.py
In Collection: ResNeSt
Metadata:
Architecture: *id001
Training Data: COCO
Name: topdown_heatmap_resnest269_coco_384x288
Results:
- Dataset: COCO
Metrics:
AP: 0.755
AP@0.5: 0.908
AP@0.75: 0.828
AR: 0.806
AR@0.5: 0.943
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/top_down/resnest/resnest269_coco_384x288-b142b9fb_20210517.pth
configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnet_coco.md 0 → 100755
View file @ 37c8cebc
<!-- [ALGORITHM] -->
<details>
<summary align="right"><a href="http://openaccess.thecvf.com/content_ECCV_2018/html/Bin_Xiao_Simple_Baselines_for_ECCV_2018_paper.html">SimpleBaseline2D (ECCV'2018)</a></summary>
```bibtex
@inproceedings{xiao2018simple,
title={Simple baselines for human pose estimation and tracking},
author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
booktitle={Proceedings of the European conference on computer vision (ECCV)},
pages={466--481},
year={2018}
}
```
</details>
<!-- [BACKBONE] -->
<details>
<summary align="right"><a href="http://openaccess.thecvf.com/content_cvpr_2016/html/He_Deep_Residual_Learning_CVPR_2016_paper.html">ResNet (CVPR'2016)</a></summary>
```bibtex
@inproceedings{he2016deep,
title={Deep residual learning for image recognition},
author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
pages={770--778},
year={2016}
}
```
</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 with detector having human AP of 56.4 on COCO val2017 dataset
| Arch | Input Size | AP | AP<sup>50</sup> | AP<sup>75</sup> | AR | AR<sup>50</sup> | ckpt | log |
| :-------------------------------------------- | :--------: | :---: | :-------------: | :-------------: | :---: | :-------------: | :-------------------------------------------: | :-------------------------------------------: |
| [pose_resnet_50](/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res50_coco_256x192.py) | 256x192 | 0.718 | 0.898 | 0.795 | 0.773 | 0.937 | [ckpt](https://download.openmmlab.com/mmpose/top_down/resnet/res50_coco_256x192-ec54d7f3_20200709.pth) | [log](https://download.openmmlab.com/mmpose/top_down/resnet/res50_coco_256x192_20200709.log.json) |
| [pose_resnet_50](/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res50_coco_384x288.py) | 384x288 | 0.731 | 0.900 | 0.799 | 0.783 | 0.931 | [ckpt](https://download.openmmlab.com/mmpose/top_down/resnet/res50_coco_384x288-e6f795e9_20200709.pth) | [log](https://download.openmmlab.com/mmpose/top_down/resnet/res50_coco_384x288_20200709.log.json) |
| [pose_resnet_101](/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res101_coco_256x192.py) | 256x192 | 0.726 | 0.899 | 0.806 | 0.781 | 0.939 | [ckpt](https://download.openmmlab.com/mmpose/top_down/resnet/res101_coco_256x192-6e6babf0_20200708.pth) | [log](https://download.openmmlab.com/mmpose/top_down/resnet/res101_coco_256x192_20200708.log.json) |
| [pose_resnet_101](/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res101_coco_384x288.py) | 384x288 | 0.748 | 0.905 | 0.817 | 0.798 | 0.940 | [ckpt](https://download.openmmlab.com/mmpose/top_down/resnet/res101_coco_384x288-8c71bdc9_20200709.pth) | [log](https://download.openmmlab.com/mmpose/top_down/resnet/res101_coco_384x288_20200709.log.json) |
| [pose_resnet_152](/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res152_coco_256x192.py) | 256x192 | 0.735 | 0.905 | 0.812 | 0.790 | 0.943 | [ckpt](https://download.openmmlab.com/mmpose/top_down/resnet/res152_coco_256x192-f6e307c2_20200709.pth) | [log](https://download.openmmlab.com/mmpose/top_down/resnet/res152_coco_256x192_20200709.log.json) |
| [pose_resnet_152](/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res152_coco_384x288.py) | 384x288 | 0.750 | 0.908 | 0.821 | 0.800 | 0.942 | [ckpt](https://download.openmmlab.com/mmpose/top_down/resnet/res152_coco_384x288-3860d4c9_20200709.pth) | [log](https://download.openmmlab.com/mmpose/top_down/resnet/res152_coco_384x288_20200709.log.json) |
configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnet_coco.yml 0 → 100755
View file @ 37c8cebc
Collections:
- Name: SimpleBaseline2D
Paper:
Title: Simple baselines for human pose estimation and tracking
URL: http://openaccess.thecvf.com/content_ECCV_2018/html/Bin_Xiao_Simple_Baselines_for_ECCV_2018_paper.html
README: https://github.com/open-mmlab/mmpose/blob/master/docs/en/papers/algorithms/simplebaseline2d.md
Models:
- Config: configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res50_coco_256x192.py
In Collection: SimpleBaseline2D
Metadata:
Architecture: &id001
- SimpleBaseline2D
- ResNet
Training Data: COCO
Name: topdown_heatmap_res50_coco_256x192
Results:
- Dataset: COCO
Metrics:
AP: 0.718
AP@0.5: 0.898
AP@0.75: 0.795
AR: 0.773
AR@0.5: 0.937
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/top_down/resnet/res50_coco_256x192-ec54d7f3_20200709.pth
- Config: configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res50_coco_384x288.py
In Collection: SimpleBaseline2D
Metadata:
Architecture: *id001
Training Data: COCO
Name: topdown_heatmap_res50_coco_384x288
Results:
- Dataset: COCO
Metrics:
AP: 0.731
AP@0.5: 0.9
AP@0.75: 0.799
AR: 0.783
AR@0.5: 0.931
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/top_down/resnet/res50_coco_384x288-e6f795e9_20200709.pth
- Config: configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res101_coco_256x192.py
In Collection: SimpleBaseline2D
Metadata:
Architecture: *id001
Training Data: COCO
Name: topdown_heatmap_res101_coco_256x192
Results:
- Dataset: COCO
Metrics:
AP: 0.726
AP@0.5: 0.899
AP@0.75: 0.806
AR: 0.781
AR@0.5: 0.939
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/top_down/resnet/res101_coco_256x192-6e6babf0_20200708.pth
- Config: configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res101_coco_384x288.py
In Collection: SimpleBaseline2D
Metadata:
Architecture: *id001
Training Data: COCO
Name: topdown_heatmap_res101_coco_384x288
Results:
- Dataset: COCO
Metrics:
AP: 0.748
AP@0.5: 0.905
AP@0.75: 0.817
AR: 0.798
AR@0.5: 0.94
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/top_down/resnet/res101_coco_384x288-8c71bdc9_20200709.pth
- Config: configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res152_coco_256x192.py
In Collection: SimpleBaseline2D
Metadata:
Architecture: *id001
Training Data: COCO
Name: topdown_heatmap_res152_coco_256x192
Results:
- Dataset: COCO
Metrics:
AP: 0.735
AP@0.5: 0.905
AP@0.75: 0.812
AR: 0.79
AR@0.5: 0.943
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/top_down/resnet/res152_coco_256x192-f6e307c2_20200709.pth
- Config: configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res152_coco_384x288.py
In Collection: SimpleBaseline2D
Metadata:
Architecture: *id001
Training Data: COCO
Name: topdown_heatmap_res152_coco_384x288
Results:
- Dataset: COCO
Metrics:
AP: 0.75
AP@0.5: 0.908
AP@0.75: 0.821
AR: 0.8
AR@0.5: 0.942
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/top_down/resnet/res152_coco_384x288-3860d4c9_20200709.pth
configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnet_dark_coco.md 0 → 100755
View file @ 37c8cebc
<!-- [ALGORITHM] -->
<details>
<summary align="right"><a href="http://openaccess.thecvf.com/content_ECCV_2018/html/Bin_Xiao_Simple_Baselines_for_ECCV_2018_paper.html">SimpleBaseline2D (ECCV'2018)</a></summary>
```bibtex
@inproceedings{xiao2018simple,
title={Simple baselines for human pose estimation and tracking},
author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
booktitle={Proceedings of the European conference on computer vision (ECCV)},
pages={466--481},
year={2018}
}
```
</details>
<!-- [BACKBONE] -->
<details>
<summary align="right"><a href="http://openaccess.thecvf.com/content_cvpr_2016/html/He_Deep_Residual_Learning_CVPR_2016_paper.html">ResNet (CVPR'2016)</a></summary>
```bibtex
@inproceedings{he2016deep,
title={Deep residual learning for image recognition},
author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
pages={770--778},
year={2016}
}
```
</details>
<!-- [ALGORITHM] -->
<details>
<summary align="right"><a href="http://openaccess.thecvf.com/content_CVPR_2020/html/Zhang_Distribution-Aware_Coordinate_Representation_for_Human_Pose_Estimation_CVPR_2020_paper.html">DarkPose (CVPR'2020)</a></summary>
```bibtex
@inproceedings{zhang2020distribution,
title={Distribution-aware coordinate representation for human pose estimation},
author={Zhang, Feng and Zhu, Xiatian and Dai, Hanbin and Ye, Mao and Zhu, Ce},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
pages={7093--7102},
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 with detector having human AP of 56.4 on COCO val2017 dataset
| Arch | Input Size | AP | AP<sup>50</sup> | AP<sup>75</sup> | AR | AR<sup>50</sup> | ckpt | log |
| :-------------------------------------------- | :--------: | :---: | :-------------: | :-------------: | :---: | :-------------: | :-------------------------------------------: | :-------------------------------------------: |
| [pose_resnet_50_dark](/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res50_coco_256x192_dark.py) | 256x192 | 0.724 | 0.898 | 0.800 | 0.777 | 0.936 | [ckpt](https://download.openmmlab.com/mmpose/top_down/resnet/res50_coco_256x192_dark-43379d20_20200709.pth) | [log](https://download.openmmlab.com/mmpose/top_down/resnet/res50_coco_256x192_dark_20200709.log.json) |
| [pose_resnet_50_dark](/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res50_coco_384x288_dark.py) | 384x288 | 0.735 | 0.900 | 0.801 | 0.785 | 0.937 | [ckpt](https://download.openmmlab.com/mmpose/top_down/resnet/res50_coco_384x288_dark-33d3e5e5_20210203.pth) | [log](https://download.openmmlab.com/mmpose/top_down/resnet/res50_coco_384x288_dark_20210203.log.json) |
| [pose_resnet_101_dark](/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res101_coco_256x192_dark.py) | 256x192 | 0.732 | 0.899 | 0.808 | 0.786 | 0.938 | [ckpt](https://download.openmmlab.com/mmpose/top_down/resnet/res101_coco_256x192_dark-64d433e6_20200812.pth) | [log](https://download.openmmlab.com/mmpose/top_down/resnet/res101_coco_256x192_dark_20200812.log.json) |
| [pose_resnet_101_dark](/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res101_coco_384x288_dark.py) | 384x288 | 0.749 | 0.902 | 0.816 | 0.799 | 0.939 | [ckpt](https://download.openmmlab.com/mmpose/top_down/resnet/res101_coco_384x288_dark-cb45c88d_20210203.pth) | [log](https://download.openmmlab.com/mmpose/top_down/resnet/res101_coco_384x288_dark_20210203.log.json) |
| [pose_resnet_152_dark](/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res152_coco_256x192_dark.py) | 256x192 | 0.745 | 0.905 | 0.821 | 0.797 | 0.942 | [ckpt](https://download.openmmlab.com/mmpose/top_down/resnet/res152_coco_256x192_dark-ab4840d5_20200812.pth) | [log](https://download.openmmlab.com/mmpose/top_down/resnet/res152_coco_256x192_dark_20200812.log.json) |
| [pose_resnet_152_dark](/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res152_coco_384x288_dark.py) | 384x288 | 0.757 | 0.909 | 0.826 | 0.806 | 0.943 | [ckpt](https://download.openmmlab.com/mmpose/top_down/resnet/res152_coco_384x288_dark-d3b8ebd7_20210203.pth) | [log](https://download.openmmlab.com/mmpose/top_down/resnet/res152_coco_384x288_dark_20210203.log.json) |
configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnet_dark_coco.yml 0 → 100755
View file @ 37c8cebc
Collections:
- Name: DarkPose
Paper:
Title: Distribution-aware coordinate representation for human pose estimation
URL: http://openaccess.thecvf.com/content_CVPR_2020/html/Zhang_Distribution-Aware_Coordinate_Representation_for_Human_Pose_Estimation_CVPR_2020_paper.html
README: https://github.com/open-mmlab/mmpose/blob/master/docs/en/papers/techniques/dark.md
Models:
- Config: configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res50_coco_256x192_dark.py
In Collection: DarkPose
Metadata:
Architecture: &id001
- SimpleBaseline2D
- ResNet
- DarkPose
Training Data: COCO
Name: topdown_heatmap_res50_coco_256x192_dark
Results:
- Dataset: COCO
Metrics:
AP: 0.724
AP@0.5: 0.898
AP@0.75: 0.8
AR: 0.777
AR@0.5: 0.936
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/top_down/resnet/res50_coco_256x192_dark-43379d20_20200709.pth
- Config: configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res50_coco_384x288_dark.py
In Collection: DarkPose
Metadata:
Architecture: *id001
Training Data: COCO
Name: topdown_heatmap_res50_coco_384x288_dark
Results:
- Dataset: COCO
Metrics:
AP: 0.735
AP@0.5: 0.9
AP@0.75: 0.801
AR: 0.785
AR@0.5: 0.937
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/top_down/resnet/res50_coco_384x288_dark-33d3e5e5_20210203.pth
- Config: configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res101_coco_256x192_dark.py
In Collection: DarkPose
Metadata:
Architecture: *id001
Training Data: COCO
Name: topdown_heatmap_res101_coco_256x192_dark
Results:
- Dataset: COCO
Metrics:
AP: 0.732
AP@0.5: 0.899
AP@0.75: 0.808
AR: 0.786
AR@0.5: 0.938
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/top_down/resnet/res101_coco_256x192_dark-64d433e6_20200812.pth
- Config: configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res101_coco_384x288_dark.py
In Collection: DarkPose
Metadata:
Architecture: *id001
Training Data: COCO
Name: topdown_heatmap_res101_coco_384x288_dark
Results:
- Dataset: COCO
Metrics:
AP: 0.749
AP@0.5: 0.902
AP@0.75: 0.816
AR: 0.799
AR@0.5: 0.939
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/top_down/resnet/res101_coco_384x288_dark-cb45c88d_20210203.pth
- Config: configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res152_coco_256x192_dark.py
In Collection: DarkPose
Metadata:
Architecture: *id001
Training Data: COCO
Name: topdown_heatmap_res152_coco_256x192_dark
Results:
- Dataset: COCO
Metrics:
AP: 0.745
AP@0.5: 0.905
AP@0.75: 0.821
AR: 0.797
AR@0.5: 0.942
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/top_down/resnet/res152_coco_256x192_dark-ab4840d5_20200812.pth
- Config: configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res152_coco_384x288_dark.py
In Collection: DarkPose
Metadata:
Architecture: *id001
Training Data: COCO
Name: topdown_heatmap_res152_coco_384x288_dark
Results:
- Dataset: COCO
Metrics:
AP: 0.757
AP@0.5: 0.909
AP@0.75: 0.826
AR: 0.806
AR@0.5: 0.943
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/top_down/resnet/res152_coco_384x288_dark-d3b8ebd7_20210203.pth
configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnet_fp16_coco.md 0 → 100755
View file @ 37c8cebc
<!-- [ALGORITHM] -->
<details>
<summary align="right"><a href="http://openaccess.thecvf.com/content_ECCV_2018/html/Bin_Xiao_Simple_Baselines_for_ECCV_2018_paper.html">SimpleBaseline2D (ECCV'2018)</a></summary>
```bibtex
@inproceedings{xiao2018simple,
title={Simple baselines for human pose estimation and tracking},
author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
booktitle={Proceedings of the European conference on computer vision (ECCV)},
pages={466--481},
year={2018}
}
```
</details>
<!-- [BACKBONE] -->
<details>
<summary align="right"><a href="http://openaccess.thecvf.com/content_cvpr_2016/html/He_Deep_Residual_Learning_CVPR_2016_paper.html">ResNet (CVPR'2016)</a></summary>
```bibtex
@inproceedings{he2016deep,
title={Deep residual learning for image recognition},
author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
pages={770--778},
year={2016}
}
```
</details>
<!-- [OTHERS] -->
<details>
<summary align="right"><a href="https://arxiv.org/abs/1710.03740">FP16 (ArXiv'2017)</a></summary>
```bibtex
@article{micikevicius2017mixed,
title={Mixed precision training},
author={Micikevicius, Paulius and Narang, Sharan and Alben, Jonah and Diamos, Gregory and Elsen, Erich and Garcia, David and Ginsburg, Boris and Houston, Michael and Kuchaiev, Oleksii and Venkatesh, Ganesh and others},
journal={arXiv preprint arXiv:1710.03740},
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 with detector having human AP of 56.4 on COCO val2017 dataset
| Arch | Input Size | AP | AP<sup>50</sup> | AP<sup>75</sup> | AR | AR<sup>50</sup> | ckpt | log |
| :-------------------------------------------- | :--------: | :---: | :-------------: | :-------------: | :---: | :-------------: | :-------------------------------------------: | :-------------------------------------------: |
| [pose_resnet_50_fp16](/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res50_coco_256x192_fp16_dynamic.py) | 256x192 | 0.717 | 0.898 | 0.793 | 0.772 | 0.936 | [ckpt](https://download.openmmlab.com/mmpose/top_down/resnet/res50_coco_256x192_fp16_dynamic-6edb79f3_20210430.pth) | [log](https://download.openmmlab.com/mmpose/top_down/resnet/res50_coco_256x192_fp16_dynamic_20210430.log.json) |
configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnet_fp16_coco.yml 0 → 100755
View file @ 37c8cebc
Collections:
- Name: SimpleBaseline2D
Paper:
Title: Simple baselines for human pose estimation and tracking
URL: http://openaccess.thecvf.com/content_ECCV_2018/html/Bin_Xiao_Simple_Baselines_for_ECCV_2018_paper.html
README: https://github.com/open-mmlab/mmpose/blob/master/docs/en/papers/algorithms/simplebaseline2d.md
Models:
- Config: configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/res50_coco_256x192_fp16_dynamic.py
In Collection: SimpleBaseline2D
Metadata:
Architecture:
- SimpleBaseline2D
- ResNet
Training Data: COCO
Name: topdown_heatmap_res50_coco_256x192_fp16_dynamic
Results:
- Dataset: COCO
Metrics:
AP: 0.717
AP@0.5: 0.898
AP@0.75: 0.793
AR: 0.772
AR@0.5: 0.936
Task: Body 2D Keypoint
Weights: https://download.openmmlab.com/mmpose/top_down/resnet/res50_coco_256x192_fp16_dynamic-6edb79f3_20210430.pth
configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/resnetv1d101_coco_256x192.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
evaluation = dict(interval=10, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=5e-4,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[170, 200])
total_epochs = 210
channel_cfg = dict(
num_output_channels=17,
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
])
# model settings
model = dict(
type='TopDown',
pretrained='mmcls://resnet101_v1d',
backbone=dict(type='ResNetV1d', depth=101),
keypoint_head=dict(
type='TopdownHeatmapSimpleHead',
in_channels=2048,
out_channels=channel_cfg['num_output_channels'],
loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
train_cfg=dict(),
test_cfg=dict(
flip_test=True,
post_process='default',
shift_heatmap=True,
modulate_kernel=11))
data_cfg = dict(
image_size=[192, 256],
heatmap_size=[48, 64],
num_output_channels=channel_cfg['num_output_channels'],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
soft_nms=False,
nms_thr=1.0,
oks_thr=0.9,
vis_thr=0.2,
use_gt_bbox=False,
det_bbox_thr=0.0,
bbox_file='data/coco/person_detection_results/'
'COCO_val2017_detections_AP_H_56_person.json',
)
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownRandomShiftBboxCenter', shift_factor=0.16, prob=0.3),
dict(type='TopDownRandomFlip', flip_prob=0.5),
dict(
type='TopDownHalfBodyTransform',
num_joints_half_body=8,
prob_half_body=0.3),
dict(
type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
dict(type='TopDownAffine'),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(type='TopDownGenerateTarget', sigma=2),
dict(
type='Collect',
keys=['img', 'target', 'target_weight'],
meta_keys=[
'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
'rotation', 'bbox_score', 'flip_pairs'
]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownAffine'),
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', 'center', 'scale', 'rotation', 'bbox_score',
'flip_pairs'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
samples_per_gpu=64,
workers_per_gpu=2,
val_dataloader=dict(samples_per_gpu=32),
test_dataloader=dict(samples_per_gpu=32),
train=dict(
type='TopDownCocoDataset',
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='TopDownCocoDataset',
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='TopDownCocoDataset',
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/topdown_heatmap/coco/resnetv1d101_coco_384x288.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
evaluation = dict(interval=10, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=5e-4,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[170, 200])
total_epochs = 210
channel_cfg = dict(
num_output_channels=17,
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
])
# model settings
model = dict(
type='TopDown',
pretrained='mmcls://resnet101_v1d',
backbone=dict(type='ResNetV1d', depth=101),
keypoint_head=dict(
type='TopdownHeatmapSimpleHead',
in_channels=2048,
out_channels=channel_cfg['num_output_channels'],
loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
train_cfg=dict(),
test_cfg=dict(
flip_test=True,
post_process='default',
shift_heatmap=True,
modulate_kernel=11))
data_cfg = dict(
image_size=[288, 384],
heatmap_size=[72, 96],
num_output_channels=channel_cfg['num_output_channels'],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
soft_nms=False,
nms_thr=1.0,
oks_thr=0.9,
vis_thr=0.2,
use_gt_bbox=False,
det_bbox_thr=0.0,
bbox_file='data/coco/person_detection_results/'
'COCO_val2017_detections_AP_H_56_person.json',
)
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownRandomShiftBboxCenter', shift_factor=0.16, prob=0.3),
dict(type='TopDownRandomFlip', flip_prob=0.5),
dict(
type='TopDownHalfBodyTransform',
num_joints_half_body=8,
prob_half_body=0.3),
dict(
type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
dict(type='TopDownAffine'),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(type='TopDownGenerateTarget', sigma=3),
dict(
type='Collect',
keys=['img', 'target', 'target_weight'],
meta_keys=[
'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
'rotation', 'bbox_score', 'flip_pairs'
]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownAffine'),
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', 'center', 'scale', 'rotation', 'bbox_score',
'flip_pairs'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
samples_per_gpu=32,
workers_per_gpu=2,
val_dataloader=dict(samples_per_gpu=32),
test_dataloader=dict(samples_per_gpu=32),
train=dict(
type='TopDownCocoDataset',
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='TopDownCocoDataset',
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='TopDownCocoDataset',
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/topdown_heatmap/coco/resnetv1d152_coco_256x192.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
evaluation = dict(interval=10, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=5e-4,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[170, 200])
total_epochs = 210
channel_cfg = dict(
num_output_channels=17,
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
])
# model settings
model = dict(
type='TopDown',
pretrained='mmcls://resnet152_v1d',
backbone=dict(type='ResNetV1d', depth=152),
keypoint_head=dict(
type='TopdownHeatmapSimpleHead',
in_channels=2048,
out_channels=channel_cfg['num_output_channels'],
loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
train_cfg=dict(),
test_cfg=dict(
flip_test=True,
post_process='default',
shift_heatmap=True,
modulate_kernel=11))
data_cfg = dict(
image_size=[192, 256],
heatmap_size=[48, 64],
num_output_channels=channel_cfg['num_output_channels'],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
soft_nms=False,
nms_thr=1.0,
oks_thr=0.9,
vis_thr=0.2,
use_gt_bbox=False,
det_bbox_thr=0.0,
bbox_file='data/coco/person_detection_results/'
'COCO_val2017_detections_AP_H_56_person.json',
)
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownRandomShiftBboxCenter', shift_factor=0.16, prob=0.3),
dict(type='TopDownRandomFlip', flip_prob=0.5),
dict(
type='TopDownHalfBodyTransform',
num_joints_half_body=8,
prob_half_body=0.3),
dict(
type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
dict(type='TopDownAffine'),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(type='TopDownGenerateTarget', sigma=2),
dict(
type='Collect',
keys=['img', 'target', 'target_weight'],
meta_keys=[
'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
'rotation', 'bbox_score', 'flip_pairs'
]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownAffine'),
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', 'center', 'scale', 'rotation', 'bbox_score',
'flip_pairs'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
samples_per_gpu=32,
workers_per_gpu=2,
val_dataloader=dict(samples_per_gpu=32),
test_dataloader=dict(samples_per_gpu=32),
train=dict(
type='TopDownCocoDataset',
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='TopDownCocoDataset',
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='TopDownCocoDataset',
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/topdown_heatmap/coco/resnetv1d152_coco_384x288.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
evaluation = dict(interval=10, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=5e-4,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[170, 200])
total_epochs = 210
channel_cfg = dict(
num_output_channels=17,
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
])
# model settings
model = dict(
type='TopDown',
pretrained='mmcls://resnet152_v1d',
backbone=dict(type='ResNetV1d', depth=152),
keypoint_head=dict(
type='TopdownHeatmapSimpleHead',
in_channels=2048,
out_channels=channel_cfg['num_output_channels'],
loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
train_cfg=dict(),
test_cfg=dict(
flip_test=True,
post_process='default',
shift_heatmap=True,
modulate_kernel=11))
data_cfg = dict(
image_size=[288, 384],
heatmap_size=[72, 96],
num_output_channels=channel_cfg['num_output_channels'],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
soft_nms=False,
nms_thr=1.0,
oks_thr=0.9,
vis_thr=0.2,
use_gt_bbox=False,
det_bbox_thr=0.0,
bbox_file='data/coco/person_detection_results/'
'COCO_val2017_detections_AP_H_56_person.json',
)
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownRandomShiftBboxCenter', shift_factor=0.16, prob=0.3),
dict(type='TopDownRandomFlip', flip_prob=0.5),
dict(
type='TopDownHalfBodyTransform',
num_joints_half_body=8,
prob_half_body=0.3),
dict(
type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
dict(type='TopDownAffine'),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(type='TopDownGenerateTarget', sigma=3),
dict(
type='Collect',
keys=['img', 'target', 'target_weight'],
meta_keys=[
'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
'rotation', 'bbox_score', 'flip_pairs'
]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownAffine'),
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', 'center', 'scale', 'rotation', 'bbox_score',
'flip_pairs'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
samples_per_gpu=48,
workers_per_gpu=2,
val_dataloader=dict(samples_per_gpu=32),
test_dataloader=dict(samples_per_gpu=32),
train=dict(
type='TopDownCocoDataset',
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='TopDownCocoDataset',
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='TopDownCocoDataset',
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/topdown_heatmap/coco/resnetv1d50_coco_256x192.py 0 → 100755
View file @ 37c8cebc
_base_ = [
'../../../../_base_/default_runtime.py',
'../../../../_base_/datasets/coco.py'
]
evaluation = dict(interval=10, metric='mAP', save_best='AP')
optimizer = dict(
type='Adam',
lr=5e-4,
)
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(
policy='step',
warmup='linear',
warmup_iters=500,
warmup_ratio=0.001,
step=[170, 200])
total_epochs = 210
channel_cfg = dict(
num_output_channels=17,
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
])
# model settings
model = dict(
type='TopDown',
pretrained='mmcls://resnet50_v1d',
backbone=dict(type='ResNetV1d', depth=50),
keypoint_head=dict(
type='TopdownHeatmapSimpleHead',
in_channels=2048,
out_channels=channel_cfg['num_output_channels'],
loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
train_cfg=dict(),
test_cfg=dict(
flip_test=True,
post_process='default',
shift_heatmap=True,
modulate_kernel=11))
data_cfg = dict(
image_size=[192, 256],
heatmap_size=[48, 64],
num_output_channels=channel_cfg['num_output_channels'],
num_joints=channel_cfg['dataset_joints'],
dataset_channel=channel_cfg['dataset_channel'],
inference_channel=channel_cfg['inference_channel'],
soft_nms=False,
nms_thr=1.0,
oks_thr=0.9,
vis_thr=0.2,
use_gt_bbox=False,
det_bbox_thr=0.0,
bbox_file='data/coco/person_detection_results/'
'COCO_val2017_detections_AP_H_56_person.json',
)
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownRandomShiftBboxCenter', shift_factor=0.16, prob=0.3),
dict(type='TopDownRandomFlip', flip_prob=0.5),
dict(
type='TopDownHalfBodyTransform',
num_joints_half_body=8,
prob_half_body=0.3),
dict(
type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
dict(type='TopDownAffine'),
dict(type='ToTensor'),
dict(
type='NormalizeTensor',
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]),
dict(type='TopDownGenerateTarget', sigma=2),
dict(
type='Collect',
keys=['img', 'target', 'target_weight'],
meta_keys=[
'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
'rotation', 'bbox_score', 'flip_pairs'
]),
]
val_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='TopDownGetBboxCenterScale', padding=1.25),
dict(type='TopDownAffine'),
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', 'center', 'scale', 'rotation', 'bbox_score',
'flip_pairs'
]),
]
test_pipeline = val_pipeline
data_root = 'data/coco'
data = dict(
samples_per_gpu=64,
workers_per_gpu=2,
val_dataloader=dict(samples_per_gpu=32),
test_dataloader=dict(samples_per_gpu=32),
train=dict(
type='TopDownCocoDataset',
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='TopDownCocoDataset',
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='TopDownCocoDataset',
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