chaowentao
commented
3 years ago Motivation: Verify the effect of the bottom-up model(such as higherhrnet) on the AIC dataset. I have tried it, but the result is not good. mAP only 0.2. So please help me, thanks.
Configs: higher_hrnet32_aic_512x512.py
Datasets: bottom_up_aic.py
Details:
higher_hrnet32_aic_512x512.py
log_level = 'INFO'
load_from = None
resume_from = None
dist_params = dict(backend='nccl')
workflow = [('train', 1)]
checkpoint_config = dict(interval=10)
evaluation = dict(interval=20, metric='mAP', key_indicator='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 = 100 # 300
log_config = dict(
interval=50,
hooks=[
dict(type='TextLoggerHook'),
# dict(type='TensorboardLoggerHook')
])
channel_cfg = dict(
num_output_channels=14,
dataset_joints=14,
dataset_channel=[
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13],
],
inference_channel=[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13])
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='BottomUp',
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='BottomUpHigherResolutionHead',
in_channels=32,
num_joints=14,
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]),
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],
with_ae=[True, False],
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),
loss_pose=dict(
type='MultiLossFactory',
num_joints=14,
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_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/aic'
data = dict(
samples_per_gpu=4, # 24
workers_per_gpu=2,
train=dict(
type='BottomUpAicDataset',
ann_file=f'{data_root}/annotations/aic_train.json',
img_prefix=f'{data_root}/ai_challenger_keypoint_train_20170902/'
'keypoint_train_images_20170902/',
data_cfg=data_cfg,
pipeline=train_pipeline),
val=dict(
type='BottomUpAicDataset',
ann_file=f'{data_root}/annotations/aic_val.json',
img_prefix=f'{data_root}/ai_challenger_keypoint_validation_20170911/'
'keypoint_validation_images_20170911/',
data_cfg=data_cfg,
pipeline=val_pipeline),
test=dict(
type='BottomUpAicDataset',
ann_file=f'{data_root}/annotations/aic_val.json',
img_prefix=f'{data_root}/ai_challenger_keypoint_validation_20170911/'
'keypoint_validation_images_20170911/',
data_cfg=data_cfg,
pipeline=val_pipeline),
)bottom_up_aic.py
import numpy as np
from xtcocotools.coco import COCO
from xtcocotools.cocoeval import COCOeval
import xtcocotools
from mmpose.datasets.builder import DATASETS
from .bottom_up_coco import BottomUpCocoDataset
@DATASETS.register_module()
class BottomUpAicDataset(BottomUpCocoDataset):
"""Aic dataset for bottom-up pose estimation.
`AI Challenger : A Large-scale Dataset for Going Deeper
in Image Understanding <https://arxiv.org/abs/1711.06475>`__
The dataset loads raw features and apply specified transforms
to return a dict containing the image tensors and other information.
AIC keypoint indexes::
0: "right_shoulder",
1: "right_elbow",
2: "right_wrist",
3: "left_shoulder",
4: "left_elbow",
5: "left_wrist",
6: "right_hip",
7: "right_knee",
8: "right_ankle",
9: "left_hip",
10: "left_knee",
11: "left_ankle",
12: "head_top",
13: "neck"
Args:
ann_file (str): Path to the annotation file.
img_prefix (str): Path to a directory where images are held.
Default: None.
data_cfg (dict): config
pipeline (list[dict | callable]): A sequence of data transforms.
test_mode (bool): Store True when building test or
validation dataset. Default: False.
"""
def __init__(self,
ann_file,
img_prefix,
data_cfg,
pipeline,
test_mode=False):
super(BottomUpCocoDataset, self).__init__(
ann_file, img_prefix, data_cfg, pipeline, test_mode=test_mode)
self.ann_info['flip_index'] = [
3, 4, 5, 0, 1, 2, 9, 10, 11, 6, 7, 8, 12, 13
]
self.ann_info['use_different_joint_weights'] = False
self.ann_info['joint_weights'] = np.array(
[1., 1.2, 1.5, 1., 1.2, 1.5, 1., 1.2, 1.5, 1., 1.2, 1.5, 1., 1.],
dtype=np.float32).reshape((self.ann_info['num_joints'], 1))
self.sigmas = np.array([
0.01388152, 0.01515228, 0.01057665, 0.01417709, 0.01497891,
0.01402144, 0.03909642, 0.03686941, 0.01981803, 0.03843971,
0.03412318, 0.02415081, 0.01291456, 0.01236173
]) * 2
self.coco = COCO(ann_file)
cats = [
cat['name'] for cat in self.coco.loadCats(self.coco.getCatIds())
]
self.classes = ['__background__'] + cats
self.num_classes = len(self.classes)
self._class_to_ind = dict(zip(self.classes, range(self.num_classes)))
self._class_to_coco_ind = dict(zip(cats, self.coco.getCatIds()))
self._coco_ind_to_class_ind = dict(
(self._class_to_coco_ind[cls], self._class_to_ind[cls])
for cls in self.classes[1:])
self.img_ids = self.coco.getImgIds()
if not test_mode:
self.img_ids = [
img_id for img_id in self.img_ids
if len(self.coco.getAnnIds(imgIds=img_id, iscrowd=None)) > 0
]
self.num_images = len(self.img_ids)
self.id2name, self.name2id = self._get_mapping_id_name(self.coco.imgs)
self.dataset_name = 'aic'
print(f'=> num_images: {self.num_images}')
def _do_python_keypoint_eval(self, res_file):
"""Keypoint evaluation using COCOAPI."""
coco_det = self.coco.loadRes(res_file)
coco_eval = COCOeval(
self.coco, coco_det, 'keypoints', self.sigmas, use_area=False)
coco_eval.params.useSegm = None
coco_eval.evaluate()
coco_eval.accumulate()
coco_eval.summarize()
stats_names = [
'AP', 'AP .5', 'AP .75', 'AP (M)', 'AP (L)', 'AR', 'AR .5',
'AR .75', 'AR (M)', 'AR (L)'
]
info_str = list(zip(stats_names, coco_eval.stats))
return info_str
pablovela5620
lucasjinreal
jin-s13
nnop
From time to time, there are requests for more checkpoints. Considering that enriching the model zoo is a good thing, we collect requests for checkpoints here. Please post the detailed configs, settings, backgrounds, motivations etc. below, and others may thumbs-up 👍 the request items. We will periodically assess them and train & release them if the needs are high.
Happy Research!