Skip to content

Commit

Permalink
add example cache file
Browse files Browse the repository at this point in the history
  • Loading branch information
@pangyyyyy
pangyyyyy committed Oct 3, 2022
1 parent 92bebb9 commit 34641c6
Showing 1 changed file with 238 additions and 0 deletions.
238 changes: 238 additions & 0 deletions configs/combine/twins_svt_hmr_mix4_coco_mmpose_l1_cache.py
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,238 @@
_base_ = ['../_base_/default_runtime.py']
use_adversarial_train = True

# evaluate
evaluation = dict(metric=['pa-mpjpe', 'mpjpe'])
# optimizer
optimizer = dict(
# backbone=dict(type='Adam', lr=2.5e-4),
backbone=dict(type='Adam', lr=5e-5),
head=dict(type='Adam', lr=2.5e-4),
disc=dict(type='Adam', lr=1e-4))
optimizer_config = dict(grad_clip=None)
# learning policy
lr_config = dict(policy='Fixed', by_epoch=False)
runner = dict(type='EpochBasedRunner', max_epochs=100)

log_config = dict(
interval=50,
hooks=[
dict(type='TextLoggerHook'),
# dict(type='TensorboardLoggerHook')
])

img_res = 224

# model settings
model = dict(
type='ImageBodyModelEstimator',
# backbone=dict(
# type='PCPVT',
# arch='base',
# in_channels=3,
# out_indices=(3, ),
# qkv_bias=True,
# norm_cfg=dict(type='LN', eps=1e-06),
# norm_after_stage=[False, False, False, True],
# drop_rate=0.0,
# attn_drop_rate=0.,
# drop_path_rate=0.3,
# init_cfg=dict(
# type='Pretrained',
# prefix='backbone',
# checkpoint='data/checkpoints/twins-pcpvt-base_3rdparty_8xb128_in1k_20220126-f8c4b0d5.pth')),
backbone=dict(
type='SVT',
arch='base',
in_channels=3,
out_indices=(3, ),
qkv_bias=True,
norm_cfg=dict(type='LN'),
norm_after_stage=[False, False, False, True],
drop_rate=0.0,
attn_drop_rate=0.,
drop_path_rate=0.3,
init_cfg=dict(
type='Pretrained',
prefix='backbone',
checkpoint='data/checkpoints/twins_svt_epoch_210.pth')),
head=dict(
type='HMRHrNetHead',
feat_dim=768,
smpl_mean_params='data/body_models/smpl_mean_params.npz'),
body_model_train=dict(
type='SMPL',
keypoint_src='smpl_54',
keypoint_dst='smpl_54',
model_path='data/body_models/smpl',
keypoint_approximate=True,
extra_joints_regressor='data/body_models/J_regressor_extra.npy'),
body_model_test=dict(
type='SMPL',
keypoint_src='h36m',
keypoint_dst='h36m',
model_path='data/body_models/smpl',
joints_regressor='data/body_models/J_regressor_h36m.npy'),
convention='smpl_54',
loss_keypoints3d=dict(type='L1Loss', loss_weight=100),
loss_keypoints2d=dict(type='L1Loss', loss_weight=10),
loss_vertex=dict(type='L1Loss', loss_weight=2),
loss_smpl_pose=dict(type='L1Loss', loss_weight=3),
loss_smpl_betas=dict(type='L1Loss', loss_weight=0.02),
loss_adv=dict(
type='GANLoss',
gan_type='lsgan',
real_label_val=1.0,
fake_label_val=0.0,
loss_weight=1),
disc=dict(type='SMPLDiscriminator'))
# dataset settings
dataset_type = 'HumanImageDataset'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
data_keys = [
'has_smpl', 'smpl_body_pose', 'smpl_global_orient', 'smpl_betas',
'smpl_transl', 'keypoints2d', 'keypoints3d', 'sample_idx'
]
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='RandomChannelNoise', noise_factor=0.4),
dict(type='RandomHorizontalFlip', flip_prob=0.5, convention='smpl_54'),
dict(type='GetRandomScaleRotation', rot_factor=30, scale_factor=0.25),
dict(type='MeshAffine', img_res=224),
dict(type='Normalize', **img_norm_cfg),
dict(type='ImageToTensor', keys=['img']),
dict(type='ToTensor', keys=data_keys),
dict(
type='Collect',
keys=['img', *data_keys],
meta_keys=['image_path', 'center', 'scale', 'rotation'])
]
adv_data_keys = [
'smpl_body_pose', 'smpl_global_orient', 'smpl_betas', 'smpl_transl'
]
train_adv_pipeline = [dict(type='Collect', keys=adv_data_keys, meta_keys=[])]
test_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='GetRandomScaleRotation', rot_factor=0, scale_factor=0),
dict(type='MeshAffine', img_res=224),
dict(type='Normalize', **img_norm_cfg),
dict(type='ImageToTensor', keys=['img']),
dict(type='ToTensor', keys=data_keys),
dict(
type='Collect',
keys=['img', *data_keys],
meta_keys=['image_path', 'center', 'scale', 'rotation'])
]

inference_pipeline = [
dict(type='MeshAffine', img_res=224),
dict(type='Normalize', **img_norm_cfg),
dict(type='ImageToTensor', keys=['img']),
dict(
type='Collect',
keys=['img', 'sample_idx'],
meta_keys=['image_path', 'center', 'scale', 'rotation'])
]

cache_files = {
'h36m': 'data/cache/h36m_mosh_train_smpl_54.npz',
'spin_mpi_inf_3dhp': 'data/cache/spin_mpi_inf_3dhp_train_smpl_54.npz',
'eft_lspet': 'data/cache/eft_lspet_train_smpl_54.npz',
'eft_mpii': 'data/cache/eft_mpii_train_smpl_54.npz',
'eft_coco': 'data/cache/eft_coco_train_smpl_54.npz'
}

data = dict(
samples_per_gpu=32,
workers_per_gpu=1,
train=dict(
type='AdversarialDataset',
train_dataset=dict(
type='MixedDataset',
configs=[
dict(
type=dataset_type,
dataset_name='h36m',
data_prefix='data',
pipeline=train_pipeline,
convention='smpl_54',
cache_data_path=cache_files['h36m'],
ann_file='h36m_mosh_train.npz'),
dict(
type=dataset_type,
dataset_name='coco',
data_prefix='data',
pipeline=train_pipeline,
convention='smpl_54',
cache_data_path=cache_files['eft_coco'],
ann_file='eft_coco_train.npz'),
dict(
type=dataset_type,
dataset_name='lspet',
data_prefix='data',
pipeline=train_pipeline,
convention='smpl_54',
cache_data_path=cache_files['eft_lspet'],
ann_file='eft_lspet_train.npz'),
dict(
type=dataset_type,
dataset_name='mpii',
data_prefix='data',
pipeline=train_pipeline,
convention='smpl_54',
cache_data_path=cache_files['eft_mpii'],
ann_file='eft_mpii_train.npz'),
# dict(
# type=dataset_type,
# dataset_name='mpii',
# data_prefix='data',
# pipeline=train_pipeline,
# convention='smpl_54',
# ann_file='mpii_train.npz'),
dict(
type=dataset_type,
dataset_name='mpi_inf_3dhp',
data_prefix='data',
pipeline=train_pipeline,
convention='smpl_54',
cache_data_path=cache_files['spin_mpi_inf_3dhp'],
ann_file='spin_mpi_inf_3dhp_train.npz'),
],
# partition=[0.35, 0.15, 0.1, 0.10, 0.10, 0.2],
partition=[0.5, 0.233, 0.046, 0.021, 0.2],
),
adv_dataset=dict(
type='MeshDataset',
dataset_name='cmu_mosh',
data_prefix='data',
pipeline=train_adv_pipeline,
ann_file='cmu_mosh.npz')),
val=dict(
type=dataset_type,
body_model=dict(
type='GenderedSMPL',
keypoint_src='h36m',
keypoint_dst='h36m',
model_path='data/body_models/smpl',
joints_regressor='data/body_models/J_regressor_h36m.npy'),
dataset_name='pw3d',
data_prefix='data',
pipeline=test_pipeline,
ann_file='pw3d_test.npz'),
test=dict(
type=dataset_type,
body_model=dict(
type='GenderedSMPL',
keypoint_src='h36m',
keypoint_dst='h36m',
model_path='data/body_models/smpl',
joints_regressor='data/body_models/J_regressor_h36m.npy'),
dataset_name='pw3d',
data_prefix='data',
pipeline=test_pipeline,
ann_file='pw3d_test.npz'),
)

custom_imports = dict(
imports=['mmhuman3d.models.backbones.twins'], allow_failed_imports=False)

0 comments on commit 34641c6

Please sign in to comment.