Tau-J
commented
1 year ago Thanks for using MMPose 1.0. For custom dataset, you should not use COCO_val2017_detections_AP_H_56_person.json, which contains bboxes detected by a detector on COCO dataset.
Closed shinysoo closed 1 year ago
Tau-J
commented
1 year ago Thanks for using MMPose 1.0. For custom dataset, you should not use COCO_val2017_detections_AP_H_56_person.json, which contains bboxes detected by a detector on COCO dataset.
shinysoo
commented
1 year ago Thanks for using MMPose 1.0. For custom dataset, you should not use
COCO_val2017_detections_AP_H_56_person.json, which contains bboxes detected by a detector on COCO dataset.
Can you share the format of COCO_val2017_detections_AP_H_56_person.json?
and how to modify the json?
Do you have any examples?
shinysoo
commented
1 year ago @Tau-J
I found the json from this Google Drive. Thanks!
But I have one question. How to calculate the score?
I have only bbox, image_id, category_id without score.
[+]
I tried with my custom COCO_val2017_detections_AP_H_56_person.json but noting changed.
(I set all the scores to 0.9.)
I think I have to change somthing more. Could you give me some advice about it? this is my config code:
_base_ = ['../../../_base_/default_runtime.py']
# runtime
train_cfg = dict(max_epochs=210, val_interval=10)
# optimizer
optim_wrapper = dict(optimizer=dict(
type='Adam',
lr=5e-4,
))
# learning policy
param_scheduler = [
dict(
type='LinearLR', begin=0, end=500, start_factor=0.001,
by_epoch=False), # warm-up
dict(
type='MultiStepLR',
begin=0,
end=210,
milestones=[170, 200],
gamma=0.1,
by_epoch=True)
]
# automatically scaling LR based on the actual training batch size
auto_scale_lr = dict(base_batch_size=512)
# hooks
default_hooks = dict(checkpoint=dict(save_best='coco/AP', rule='greater'))
# codec settings
codec = dict(
type='MSRAHeatmap', input_size=(192, 256), heatmap_size=(48, 64), sigma=2)
# model settings
model = dict(
type='TopdownPoseEstimator',
data_preprocessor=dict(
type='PoseDataPreprocessor',
mean=[123.675, 116.28, 103.53],
std=[58.395, 57.12, 57.375],
bgr_to_rgb=True),
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))),
init_cfg=dict(
type='Pretrained',
checkpoint='https://download.openmmlab.com/mmpose/'
'pretrain_models/hrnet_w32-36af842e.pth'),
),
head=dict(
type='HeatmapHead',
in_channels=32,
out_channels=17,
deconv_out_channels=None,
loss=dict(type='KeypointMSELoss', use_target_weight=True),
decoder=codec),
test_cfg=dict(
flip_test=True,
flip_mode='heatmap',
shift_heatmap=True,
))
# base dataset settings
dataset_type = 'CocoDataset'
data_mode = 'topdown'
data_root = 'data_custom/coco/' ## I modified ##
# pipelines
train_pipeline = [
dict(type='LoadImage', file_client_args={{_base_.file_client_args}}),
dict(type='GetBBoxCenterScale'),
dict(type='RandomFlip', direction='horizontal'),
dict(type='RandomHalfBody'),
dict(type='RandomBBoxTransform'),
dict(type='TopdownAffine', input_size=codec['input_size']),
dict(type='GenerateTarget', encoder=codec),
dict(type='PackPoseInputs')
]
val_pipeline = [
dict(type='LoadImage', file_client_args={{_base_.file_client_args}}),
dict(type='GetBBoxCenterScale'),
dict(type='TopdownAffine', input_size=codec['input_size']),
dict(type='PackPoseInputs')
]
# data loaders
train_dataloader = dict(
batch_size=64,
num_workers=2,
persistent_workers=True,
sampler=dict(type='DefaultSampler', shuffle=True),
dataset=dict(
type=dataset_type,
data_root=data_root,
data_mode=data_mode,
ann_file='annotations/person_keypoints_train_custom.json', ## I modified ##
data_prefix=dict(img='train_custom/'), ## I modified ##
pipeline=train_pipeline,
))
val_dataloader = dict(
batch_size=32,
num_workers=2,
persistent_workers=True,
drop_last=False,
sampler=dict(type='DefaultSampler', shuffle=False, round_up=False),
dataset=dict(
type=dataset_type,
data_root=data_root,
data_mode=data_mode,
ann_file='annotations/person_keypoints_val_custom.json',
bbox_file='data_custom/coco/person_detection_results/'
'COCO_val_detections_custom.json', ## I modified ##
data_prefix=dict(img='val_custom/'), ## I modified ##
test_mode=True,
pipeline=val_pipeline,
))
test_dataloader = val_dataloader
# evaluators
val_evaluator = dict(
type='CocoMetric',
ann_file=data_root + 'annotations/person_keypoints_val_custom.json') ## I modified ##
test_evaluator = val_evaluatorI didn't modified dataset_type = 'CocoDataset' because I've modified my annotation format to COCO format.
Tau-J
commented
1 year ago bbox_file is not necessary for custom dataset, you can just comment it out. Then the evaluation result will be based on your GT.
jin-s13
commented
1 year ago It seems the problem is solved. I will close this issue for now. Feel free to reopen it if have further questions.
Hello, I'm trying to train model with my custom datasets. so I prepared train images and annotation json files as coco style.
After that, I changed cofig file(
td-hm_hrnet-w32_8xb64-210e_coco-256x192_custom.py) is as follows:And the demo script is as follows:
Right, I've only modified the paths of annotations and images in the config file. And I compared "origin predictions result" and "custom-train predictions result".
As expected, there was no difference between "origin predictions result" and "custom-train predictions result". I think the custom config didn't work and the origin config file worked twice.
What's the problem in the process? Should I change this code?:
or should I modify the
{MMDET_CONFIG_FILE}and{~_CHECKPOINT_FILE}in the demo script?Please find the poblems in my code. Thank you.