2D hand pose by YOLOX-POSE

[ghost]

Hi, there. I wanna to train 2D hand keypoints of coco wholebody hand by YOLOX-POSE.
Is this possible??
I tried to edit this config file. But it was difficult to me.

[ghost]

[Ben-Louis]

To better assist you in resolving the issue, it would be beneficial if you could provide the complete contents of your configuration file.

[ghost]

This is config file I edited.

[Ben-Louis]

Please refer to

mmpose/configs/hand_2d_keypoint/topdown_heatmap/coco_wholebody_hand/td-hm_hrnetv2-w18_dark-8xb32-210e_coco-wholebody-hand-256x256.py

Lines 98 to 158 in 8dd07a7

	# base dataset settings
	dataset_type = 'CocoWholeBodyHandDataset'
	data_mode = 'topdown'
	data_root = 'data/coco/'
	# pipelines
	train_pipeline = [
	dict(type='LoadImage'),
	dict(type='GetBBoxCenterScale'),
	dict(
	type='RandomBBoxTransform', rotate_factor=180,
	scale_factor=(0.7, 1.3)),
	dict(type='RandomFlip', direction='horizontal'),
	dict(type='TopdownAffine', input_size=codec['input_size']),
	dict(type='GenerateTarget', encoder=codec),
	dict(type='PackPoseInputs')
	]
	val_pipeline = [
	dict(type='LoadImage'),
	dict(type='GetBBoxCenterScale'),
	dict(type='TopdownAffine', input_size=codec['input_size']),
	dict(type='PackPoseInputs')
	]
	# data loaders
	train_dataloader = dict(
	batch_size=32,
	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/coco_wholebody_train_v1.0.json',
	data_prefix=dict(img='train2017/'),
	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/coco_wholebody_val_v1.0.json',
	data_prefix=dict(img='val2017/'),
	test_mode=True,
	pipeline=val_pipeline,
	))
	test_dataloader = val_dataloader
	val_evaluator = [
	dict(type='PCKAccuracy', thr=0.2),
	dict(type='AUC'),
	dict(type='EPE')
	]
	test_evaluator = val_evaluator

to modify the arguments dataset_type and val_evaluator

[Ben-Louis]

Simultaneously, you may need to add the following code in the $MMPOSE/projects/yolox-pose/datasets/coco_dataset.py:

from mmpose.datasets import CocoWholeBodyHandDataset as MMPoseCocoWholeBodyHandDataset

@DATASETS.register_module()
class CocoWholeBodyHandDataset(MMPoseCocoWholeBodyHandDataset):

    @force_full_init
    def prepare_data(self, idx) -> Any:
        data_info = self.get_data_info(idx)
        data_info['dataset'] = self
        return self.pipeline(data_info)

[ghost]

Thank you for always.
I have to do other projects now. So I'll try them weekend.