diff --git a/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-base-simple_8xb64-210e_coco-256x192.py b/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-base-simple_8xb64-210e_coco-256x192.py
new file mode 100644
index 0000000000..b3edbfa4b2
--- /dev/null
+++ b/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-base-simple_8xb64-210e_coco-256x192.py
@@ -0,0 +1,151 @@
+_base_ = ['../../../_base_/default_runtime.py']
+
+# runtime
+train_cfg = dict(max_epochs=210, val_interval=10)
+
+# optimizer
+custom_imports = dict(
+    imports=['mmpose.engine.optim_wrappers.layer_decay_optim_wrapper'],
+    allow_failed_imports=False)
+
+optim_wrapper = dict(
+    optimizer=dict(
+        type='AdamW', lr=5e-4, betas=(0.9, 0.999), weight_decay=0.1),
+    paramwise_cfg=dict(
+        num_layers=12,
+        layer_decay_rate=0.75,
+        custom_keys={
+            'bias': dict(decay_multi=0.0),
+            'pos_embed': dict(decay_mult=0.0),
+            'relative_position_bias_table': dict(decay_mult=0.0),
+            'norm': dict(decay_mult=0.0),
+        },
+    ),
+    constructor='LayerDecayOptimWrapperConstructor',
+    clip_grad=dict(max_norm=1., norm_type=2),
+)
+
+# 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', max_keep_ckpts=1))
+
+# codec settings
+codec = dict(
+    type='UDPHeatmap', 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='mmcls.VisionTransformer',
+        arch='base',
+        img_size=(256, 192),
+        patch_size=16,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+        with_cls_token=False,
+        output_cls_token=False,
+        patch_cfg=dict(padding=2),
+        init_cfg=dict(
+            type='Pretrained',
+            checkpoint='pretrained/mae_pretrain_vit_base.pth'),
+    ),
+    head=dict(
+        type='HeatmapHead',
+        in_channels=768,
+        out_channels=17,
+        deconv_out_channels=[],
+        deconv_kernel_sizes=[],
+        loss=dict(type='KeypointMSELoss', use_target_weight=True),
+        decoder=codec,
+        extra=dict(upsample=4, final_conv_kernel=3),
+    ),
+    test_cfg=dict(
+        flip_test=True,
+        flip_mode='heatmap',
+        shift_heatmap=False,
+    ))
+
+# base dataset settings
+data_root = 'data/coco/'
+dataset_type = 'CocoDataset'
+data_mode = 'topdown'
+
+# 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'], use_udp=True),
+    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'], use_udp=True),
+    dict(type='PackPoseInputs')
+]
+
+# data loaders
+train_dataloader = dict(
+    batch_size=64,
+    num_workers=4,
+    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_train2017.json',
+        data_prefix=dict(img='train2017/'),
+        pipeline=train_pipeline,
+    ))
+val_dataloader = dict(
+    batch_size=32,
+    num_workers=4,
+    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_val2017.json',
+        bbox_file='data/coco/person_detection_results/'
+        'COCO_val2017_detections_AP_H_56_person.json',
+        data_prefix=dict(img='val2017/'),
+        test_mode=True,
+        pipeline=val_pipeline,
+    ))
+test_dataloader = val_dataloader
+
+# evaluators
+val_evaluator = dict(
+    type='CocoMetric',
+    ann_file=data_root + 'annotations/person_keypoints_val2017.json')
+test_evaluator = val_evaluator
diff --git a/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-base_8xb64-210e_coco-256x192.py b/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-base_8xb64-210e_coco-256x192.py
new file mode 100644
index 0000000000..f1fbd2d857
--- /dev/null
+++ b/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-base_8xb64-210e_coco-256x192.py
@@ -0,0 +1,149 @@
+_base_ = ['../../../_base_/default_runtime.py']
+
+# runtime
+train_cfg = dict(max_epochs=210, val_interval=10)
+
+# optimizer
+custom_imports = dict(
+    imports=['mmpose.engine.optim_wrappers.layer_decay_optim_wrapper'],
+    allow_failed_imports=False)
+
+optim_wrapper = dict(
+    optimizer=dict(
+        type='AdamW', lr=5e-4, betas=(0.9, 0.999), weight_decay=0.1),
+    paramwise_cfg=dict(
+        num_layers=12,
+        layer_decay_rate=0.75,
+        custom_keys={
+            'bias': dict(decay_multi=0.0),
+            'pos_embed': dict(decay_mult=0.0),
+            'relative_position_bias_table': dict(decay_mult=0.0),
+            'norm': dict(decay_mult=0.0),
+        },
+    ),
+    constructor='LayerDecayOptimWrapperConstructor',
+    clip_grad=dict(max_norm=1., norm_type=2),
+)
+
+# 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', max_keep_ckpts=1))
+
+# codec settings
+codec = dict(
+    type='UDPHeatmap', 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='mmcls.VisionTransformer',
+        arch='base',
+        img_size=(256, 192),
+        patch_size=16,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+        with_cls_token=False,
+        output_cls_token=False,
+        patch_cfg=dict(padding=2),
+        init_cfg=dict(
+            type='Pretrained',
+            checkpoint='pretrained/mae_pretrain_vit_base.pth'),
+    ),
+    head=dict(
+        type='HeatmapHead',
+        in_channels=768,
+        out_channels=17,
+        deconv_out_channels=(256, 256),
+        deconv_kernel_sizes=(4, 4),
+        loss=dict(type='KeypointMSELoss', use_target_weight=True),
+        decoder=codec),
+    test_cfg=dict(
+        flip_test=True,
+        flip_mode='heatmap',
+        shift_heatmap=False,
+    ))
+
+# base dataset settings
+data_root = 'data/coco/'
+dataset_type = 'CocoDataset'
+data_mode = 'topdown'
+
+# 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'], use_udp=True),
+    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'], use_udp=True),
+    dict(type='PackPoseInputs')
+]
+
+# data loaders
+train_dataloader = dict(
+    batch_size=64,
+    num_workers=4,
+    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_train2017.json',
+        data_prefix=dict(img='train2017/'),
+        pipeline=train_pipeline,
+    ))
+val_dataloader = dict(
+    batch_size=32,
+    num_workers=4,
+    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_val2017.json',
+        bbox_file='data/coco/person_detection_results/'
+        'COCO_val2017_detections_AP_H_56_person.json',
+        data_prefix=dict(img='val2017/'),
+        test_mode=True,
+        pipeline=val_pipeline,
+    ))
+test_dataloader = val_dataloader
+
+# evaluators
+val_evaluator = dict(
+    type='CocoMetric',
+    ann_file=data_root + 'annotations/person_keypoints_val2017.json')
+test_evaluator = val_evaluator
diff --git a/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-huge-simple_8xb64-210e_coco-256x192.py b/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-huge-simple_8xb64-210e_coco-256x192.py
new file mode 100644
index 0000000000..797192cb25
--- /dev/null
+++ b/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-huge-simple_8xb64-210e_coco-256x192.py
@@ -0,0 +1,151 @@
+_base_ = ['../../../_base_/default_runtime.py']
+
+# runtime
+train_cfg = dict(max_epochs=210, val_interval=10)
+
+# optimizer
+custom_imports = dict(
+    imports=['mmpose.engine.optim_wrappers.layer_decay_optim_wrapper'],
+    allow_failed_imports=False)
+
+optim_wrapper = dict(
+    optimizer=dict(
+        type='AdamW', lr=5e-4, betas=(0.9, 0.999), weight_decay=0.1),
+    paramwise_cfg=dict(
+        num_layers=32,
+        layer_decay_rate=0.85,
+        custom_keys={
+            'bias': dict(decay_multi=0.0),
+            'pos_embed': dict(decay_mult=0.0),
+            'relative_position_bias_table': dict(decay_mult=0.0),
+            'norm': dict(decay_mult=0.0),
+        },
+    ),
+    constructor='LayerDecayOptimWrapperConstructor',
+    clip_grad=dict(max_norm=1., norm_type=2),
+)
+
+# 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', max_keep_ckpts=1))
+
+# codec settings
+codec = dict(
+    type='UDPHeatmap', 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='mmcls.VisionTransformer',
+        arch='huge',
+        img_size=(256, 192),
+        patch_size=16,
+        qkv_bias=True,
+        drop_path_rate=0.55,
+        with_cls_token=False,
+        output_cls_token=False,
+        patch_cfg=dict(padding=2),
+        init_cfg=dict(
+            type='Pretrained',
+            checkpoint='pretrained/mae_pretrain_vit_huge.pth'),
+    ),
+    head=dict(
+        type='HeatmapHead',
+        in_channels=1280,
+        out_channels=17,
+        deconv_out_channels=[],
+        deconv_kernel_sizes=[],
+        loss=dict(type='KeypointMSELoss', use_target_weight=True),
+        decoder=codec,
+        extra=dict(upsample=4, final_conv_kernel=3),
+    ),
+    test_cfg=dict(
+        flip_test=True,
+        flip_mode='heatmap',
+        shift_heatmap=False,
+    ))
+
+# base dataset settings
+data_root = 'data/coco/'
+dataset_type = 'CocoDataset'
+data_mode = 'topdown'
+
+# 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'], use_udp=True),
+    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'], use_udp=True),
+    dict(type='PackPoseInputs')
+]
+
+# data loaders
+train_dataloader = dict(
+    batch_size=64,
+    num_workers=4,
+    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_train2017.json',
+        data_prefix=dict(img='train2017/'),
+        pipeline=train_pipeline,
+    ))
+val_dataloader = dict(
+    batch_size=32,
+    num_workers=4,
+    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_val2017.json',
+        bbox_file='data/coco/person_detection_results/'
+        'COCO_val2017_detections_AP_H_56_person.json',
+        data_prefix=dict(img='val2017/'),
+        test_mode=True,
+        pipeline=val_pipeline,
+    ))
+test_dataloader = val_dataloader
+
+# evaluators
+val_evaluator = dict(
+    type='CocoMetric',
+    ann_file=data_root + 'annotations/person_keypoints_val2017.json')
+test_evaluator = val_evaluator
diff --git a/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-huge_8xb64-210e_coco-256x192.py b/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-huge_8xb64-210e_coco-256x192.py
new file mode 100644
index 0000000000..43df966568
--- /dev/null
+++ b/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-huge_8xb64-210e_coco-256x192.py
@@ -0,0 +1,149 @@
+_base_ = ['../../../_base_/default_runtime.py']
+
+# runtime
+train_cfg = dict(max_epochs=210, val_interval=10)
+
+# optimizer
+custom_imports = dict(
+    imports=['mmpose.engine.optim_wrappers.layer_decay_optim_wrapper'],
+    allow_failed_imports=False)
+
+optim_wrapper = dict(
+    optimizer=dict(
+        type='AdamW', lr=5e-4, betas=(0.9, 0.999), weight_decay=0.1),
+    paramwise_cfg=dict(
+        num_layers=32,
+        layer_decay_rate=0.85,
+        custom_keys={
+            'bias': dict(decay_multi=0.0),
+            'pos_embed': dict(decay_mult=0.0),
+            'relative_position_bias_table': dict(decay_mult=0.0),
+            'norm': dict(decay_mult=0.0),
+        },
+    ),
+    constructor='LayerDecayOptimWrapperConstructor',
+    clip_grad=dict(max_norm=1., norm_type=2),
+)
+
+# 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', max_keep_ckpts=1))
+
+# codec settings
+codec = dict(
+    type='UDPHeatmap', 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='mmcls.VisionTransformer',
+        arch='huge',
+        img_size=(256, 192),
+        patch_size=16,
+        qkv_bias=True,
+        drop_path_rate=0.55,
+        with_cls_token=False,
+        output_cls_token=False,
+        patch_cfg=dict(padding=2),
+        init_cfg=dict(
+            type='Pretrained',
+            checkpoint='pretrained/mae_pretrain_vit_huge.pth'),
+    ),
+    head=dict(
+        type='HeatmapHead',
+        in_channels=1280,
+        out_channels=17,
+        deconv_out_channels=(256, 256),
+        deconv_kernel_sizes=(4, 4),
+        loss=dict(type='KeypointMSELoss', use_target_weight=True),
+        decoder=codec),
+    test_cfg=dict(
+        flip_test=True,
+        flip_mode='heatmap',
+        shift_heatmap=False,
+    ))
+
+# base dataset settings
+data_root = 'data/coco/'
+dataset_type = 'CocoDataset'
+data_mode = 'topdown'
+
+# 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'], use_udp=True),
+    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'], use_udp=True),
+    dict(type='PackPoseInputs')
+]
+
+# data loaders
+train_dataloader = dict(
+    batch_size=64,
+    num_workers=4,
+    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_train2017.json',
+        data_prefix=dict(img='train2017/'),
+        pipeline=train_pipeline,
+    ))
+val_dataloader = dict(
+    batch_size=32,
+    num_workers=4,
+    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_val2017.json',
+        bbox_file='data/coco/person_detection_results/'
+        'COCO_val2017_detections_AP_H_56_person.json',
+        data_prefix=dict(img='val2017/'),
+        test_mode=True,
+        pipeline=val_pipeline,
+    ))
+test_dataloader = val_dataloader
+
+# evaluators
+val_evaluator = dict(
+    type='CocoMetric',
+    ann_file=data_root + 'annotations/person_keypoints_val2017.json')
+test_evaluator = val_evaluator
diff --git a/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-large-simple_8xb64-210e_coco-256x192.py b/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-large-simple_8xb64-210e_coco-256x192.py
new file mode 100644
index 0000000000..9413665e6a
--- /dev/null
+++ b/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-large-simple_8xb64-210e_coco-256x192.py
@@ -0,0 +1,151 @@
+_base_ = ['../../../_base_/default_runtime.py']
+
+# runtime
+train_cfg = dict(max_epochs=210, val_interval=10)
+
+# optimizer
+custom_imports = dict(
+    imports=['mmpose.engine.optim_wrappers.layer_decay_optim_wrapper'],
+    allow_failed_imports=False)
+
+optim_wrapper = dict(
+    optimizer=dict(
+        type='AdamW', lr=5e-4, betas=(0.9, 0.999), weight_decay=0.1),
+    paramwise_cfg=dict(
+        num_layers=24,
+        layer_decay_rate=0.8,
+        custom_keys={
+            'bias': dict(decay_multi=0.0),
+            'pos_embed': dict(decay_mult=0.0),
+            'relative_position_bias_table': dict(decay_mult=0.0),
+            'norm': dict(decay_mult=0.0),
+        },
+    ),
+    constructor='LayerDecayOptimWrapperConstructor',
+    clip_grad=dict(max_norm=1., norm_type=2),
+)
+
+# 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', max_keep_ckpts=1))
+
+# codec settings
+codec = dict(
+    type='UDPHeatmap', 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='mmcls.VisionTransformer',
+        arch='large',
+        img_size=(256, 192),
+        patch_size=16,
+        qkv_bias=True,
+        drop_path_rate=0.5,
+        with_cls_token=False,
+        output_cls_token=False,
+        patch_cfg=dict(padding=2),
+        init_cfg=dict(
+            type='Pretrained',
+            checkpoint='pretrained/mae_pretrain_vit_large.pth'),
+    ),
+    head=dict(
+        type='HeatmapHead',
+        in_channels=1024,
+        out_channels=17,
+        deconv_out_channels=[],
+        deconv_kernel_sizes=[],
+        loss=dict(type='KeypointMSELoss', use_target_weight=True),
+        decoder=codec,
+        extra=dict(upsample=4, final_conv_kernel=3),
+    ),
+    test_cfg=dict(
+        flip_test=True,
+        flip_mode='heatmap',
+        shift_heatmap=False,
+    ))
+
+# base dataset settings
+data_root = 'data/coco/'
+dataset_type = 'CocoDataset'
+data_mode = 'topdown'
+
+# 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'], use_udp=True),
+    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'], use_udp=True),
+    dict(type='PackPoseInputs')
+]
+
+# data loaders
+train_dataloader = dict(
+    batch_size=64,
+    num_workers=4,
+    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_train2017.json',
+        data_prefix=dict(img='train2017/'),
+        pipeline=train_pipeline,
+    ))
+val_dataloader = dict(
+    batch_size=32,
+    num_workers=4,
+    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_val2017.json',
+        bbox_file='data/coco/person_detection_results/'
+        'COCO_val2017_detections_AP_H_56_person.json',
+        data_prefix=dict(img='val2017/'),
+        test_mode=True,
+        pipeline=val_pipeline,
+    ))
+test_dataloader = val_dataloader
+
+# evaluators
+val_evaluator = dict(
+    type='CocoMetric',
+    ann_file=data_root + 'annotations/person_keypoints_val2017.json')
+test_evaluator = val_evaluator
diff --git a/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-large_8xb64-210e_coco-256x192.py b/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-large_8xb64-210e_coco-256x192.py
new file mode 100644
index 0000000000..3f67f9999f
--- /dev/null
+++ b/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-large_8xb64-210e_coco-256x192.py
@@ -0,0 +1,149 @@
+_base_ = ['../../../_base_/default_runtime.py']
+
+# runtime
+train_cfg = dict(max_epochs=210, val_interval=10)
+
+# optimizer
+custom_imports = dict(
+    imports=['mmpose.engine.optim_wrappers.layer_decay_optim_wrapper'],
+    allow_failed_imports=False)
+
+optim_wrapper = dict(
+    optimizer=dict(
+        type='AdamW', lr=5e-4, betas=(0.9, 0.999), weight_decay=0.1),
+    paramwise_cfg=dict(
+        num_layers=24,
+        layer_decay_rate=0.8,
+        custom_keys={
+            'bias': dict(decay_multi=0.0),
+            'pos_embed': dict(decay_mult=0.0),
+            'relative_position_bias_table': dict(decay_mult=0.0),
+            'norm': dict(decay_mult=0.0),
+        },
+    ),
+    constructor='LayerDecayOptimWrapperConstructor',
+    clip_grad=dict(max_norm=1., norm_type=2),
+)
+
+# 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', max_keep_ckpts=1))
+
+# codec settings
+codec = dict(
+    type='UDPHeatmap', 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='mmcls.VisionTransformer',
+        arch='large',
+        img_size=(256, 192),
+        patch_size=16,
+        qkv_bias=True,
+        drop_path_rate=0.5,
+        with_cls_token=False,
+        output_cls_token=False,
+        patch_cfg=dict(padding=2),
+        init_cfg=dict(
+            type='Pretrained',
+            checkpoint='pretrained/mae_pretrain_vit_large.pth'),
+    ),
+    head=dict(
+        type='HeatmapHead',
+        in_channels=1024,
+        out_channels=17,
+        deconv_out_channels=(256, 256),
+        deconv_kernel_sizes=(4, 4),
+        loss=dict(type='KeypointMSELoss', use_target_weight=True),
+        decoder=codec),
+    test_cfg=dict(
+        flip_test=True,
+        flip_mode='heatmap',
+        shift_heatmap=False,
+    ))
+
+# base dataset settings
+data_root = 'data/coco/'
+dataset_type = 'CocoDataset'
+data_mode = 'topdown'
+
+# 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'], use_udp=True),
+    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'], use_udp=True),
+    dict(type='PackPoseInputs')
+]
+
+# data loaders
+train_dataloader = dict(
+    batch_size=64,
+    num_workers=4,
+    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_train2017.json',
+        data_prefix=dict(img='train2017/'),
+        pipeline=train_pipeline,
+    ))
+val_dataloader = dict(
+    batch_size=32,
+    num_workers=4,
+    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_val2017.json',
+        bbox_file='data/coco/person_detection_results/'
+        'COCO_val2017_detections_AP_H_56_person.json',
+        data_prefix=dict(img='val2017/'),
+        test_mode=True,
+        pipeline=val_pipeline,
+    ))
+test_dataloader = val_dataloader
+
+# evaluators
+val_evaluator = dict(
+    type='CocoMetric',
+    ann_file=data_root + 'annotations/person_keypoints_val2017.json')
+test_evaluator = val_evaluator
diff --git a/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-small-simple_8xb64-210e_coco-256x192.py b/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-small-simple_8xb64-210e_coco-256x192.py
new file mode 100644
index 0000000000..fdd8428891
--- /dev/null
+++ b/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-small-simple_8xb64-210e_coco-256x192.py
@@ -0,0 +1,156 @@
+_base_ = ['../../../_base_/default_runtime.py']
+
+# runtime
+train_cfg = dict(max_epochs=210, val_interval=10)
+
+# optimizer
+custom_imports = dict(
+    imports=['mmpose.engine.optim_wrappers.layer_decay_optim_wrapper'],
+    allow_failed_imports=False)
+
+optim_wrapper = dict(
+    optimizer=dict(
+        type='AdamW', lr=5e-4, betas=(0.9, 0.999), weight_decay=0.1),
+    paramwise_cfg=dict(
+        num_layers=12,
+        layer_decay_rate=0.8,
+        custom_keys={
+            'bias': dict(decay_multi=0.0),
+            'pos_embed': dict(decay_mult=0.0),
+            'relative_position_bias_table': dict(decay_mult=0.0),
+            'norm': dict(decay_mult=0.0),
+        },
+    ),
+    constructor='LayerDecayOptimWrapperConstructor',
+    clip_grad=dict(max_norm=1., norm_type=2),
+)
+
+# 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', max_keep_ckpts=1))
+
+# codec settings
+codec = dict(
+    type='UDPHeatmap', 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='mmcls.VisionTransformer',
+        arch={
+            'embed_dims': 384,
+            'num_layers': 12,
+            'num_heads': 12,
+            'feedforward_channels': 384 * 4
+        },
+        img_size=(256, 192),
+        patch_size=16,
+        qkv_bias=True,
+        drop_path_rate=0.1,
+        with_cls_token=False,
+        output_cls_token=False,
+        patch_cfg=dict(padding=2),
+        init_cfg=dict(
+            type='Pretrained',
+            checkpoint='pretrained/mae_pretrain_vit_small.pth'),
+    ),
+    head=dict(
+        type='HeatmapHead',
+        in_channels=384,
+        out_channels=17,
+        deconv_out_channels=[],
+        deconv_kernel_sizes=[],
+        loss=dict(type='KeypointMSELoss', use_target_weight=True),
+        decoder=codec,
+        extra=dict(upsample=4, final_conv_kernel=3),
+    ),
+    test_cfg=dict(
+        flip_test=True,
+        flip_mode='heatmap',
+        shift_heatmap=False,
+    ))
+
+# base dataset settings
+data_root = 'data/coco/'
+dataset_type = 'CocoDataset'
+data_mode = 'topdown'
+
+# 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'], use_udp=True),
+    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'], use_udp=True),
+    dict(type='PackPoseInputs')
+]
+
+# data loaders
+train_dataloader = dict(
+    batch_size=64,
+    num_workers=4,
+    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_train2017.json',
+        data_prefix=dict(img='train2017/'),
+        pipeline=train_pipeline,
+    ))
+val_dataloader = dict(
+    batch_size=32,
+    num_workers=4,
+    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_val2017.json',
+        bbox_file='data/coco/person_detection_results/'
+        'COCO_val2017_detections_AP_H_56_person.json',
+        data_prefix=dict(img='val2017/'),
+        test_mode=True,
+        pipeline=val_pipeline,
+    ))
+test_dataloader = val_dataloader
+
+# evaluators
+val_evaluator = dict(
+    type='CocoMetric',
+    ann_file=data_root + 'annotations/person_keypoints_val2017.json')
+test_evaluator = val_evaluator
diff --git a/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-small_8xb64-210e_coco-256x192.py b/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-small_8xb64-210e_coco-256x192.py
new file mode 100644
index 0000000000..f50ce7a9c7
--- /dev/null
+++ b/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-small_8xb64-210e_coco-256x192.py
@@ -0,0 +1,154 @@
+_base_ = ['../../../_base_/default_runtime.py']
+
+# runtime
+train_cfg = dict(max_epochs=210, val_interval=10)
+
+# optimizer
+custom_imports = dict(
+    imports=['mmpose.engine.optim_wrappers.layer_decay_optim_wrapper'],
+    allow_failed_imports=False)
+
+optim_wrapper = dict(
+    optimizer=dict(
+        type='AdamW', lr=5e-4, betas=(0.9, 0.999), weight_decay=0.1),
+    paramwise_cfg=dict(
+        num_layers=12,
+        layer_decay_rate=0.8,
+        custom_keys={
+            'bias': dict(decay_multi=0.0),
+            'pos_embed': dict(decay_mult=0.0),
+            'relative_position_bias_table': dict(decay_mult=0.0),
+            'norm': dict(decay_mult=0.0),
+        },
+    ),
+    constructor='LayerDecayOptimWrapperConstructor',
+    clip_grad=dict(max_norm=1., norm_type=2),
+)
+
+# 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', max_keep_ckpts=1))
+
+# codec settings
+codec = dict(
+    type='UDPHeatmap', 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='mmcls.VisionTransformer',
+        arch={
+            'embed_dims': 384,
+            'num_layers': 12,
+            'num_heads': 12,
+            'feedforward_channels': 384 * 4
+        },
+        img_size=(256, 192),
+        patch_size=16,
+        qkv_bias=True,
+        drop_path_rate=0.1,
+        with_cls_token=False,
+        output_cls_token=False,
+        patch_cfg=dict(padding=2),
+        init_cfg=dict(
+            type='Pretrained',
+            checkpoint='pretrained/mae_pretrain_vit_small.pth'),
+    ),
+    head=dict(
+        type='HeatmapHead',
+        in_channels=384,
+        out_channels=17,
+        deconv_out_channels=(256, 256),
+        deconv_kernel_sizes=(4, 4),
+        loss=dict(type='KeypointMSELoss', use_target_weight=True),
+        decoder=codec),
+    test_cfg=dict(
+        flip_test=True,
+        flip_mode='heatmap',
+        shift_heatmap=False,
+    ))
+
+# base dataset settings
+data_root = 'data/coco/'
+dataset_type = 'CocoDataset'
+data_mode = 'topdown'
+
+# 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'], use_udp=True),
+    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'], use_udp=True),
+    dict(type='PackPoseInputs')
+]
+
+# data loaders
+train_dataloader = dict(
+    batch_size=64,
+    num_workers=4,
+    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_train2017.json',
+        data_prefix=dict(img='train2017/'),
+        pipeline=train_pipeline,
+    ))
+val_dataloader = dict(
+    batch_size=32,
+    num_workers=4,
+    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_val2017.json',
+        bbox_file='data/coco/person_detection_results/'
+        'COCO_val2017_detections_AP_H_56_person.json',
+        data_prefix=dict(img='val2017/'),
+        test_mode=True,
+        pipeline=val_pipeline,
+    ))
+test_dataloader = val_dataloader
+
+# evaluators
+val_evaluator = dict(
+    type='CocoMetric',
+    ann_file=data_root + 'annotations/person_keypoints_val2017.json')
+test_evaluator = val_evaluator
diff --git a/configs/body_2d_keypoint/topdown_heatmap/coco/vitpose_coco.md b/configs/body_2d_keypoint/topdown_heatmap/coco/vitpose_coco.md
new file mode 100644
index 0000000000..77c1b03124
--- /dev/null
+++ b/configs/body_2d_keypoint/topdown_heatmap/coco/vitpose_coco.md
@@ -0,0 +1,58 @@
+<!-- [BACKBONE] -->
+
+<details>
+<summary align="right"><a href="https://arxiv.org/abs/2204.12484">ViTPose</a></summary>
+
+```bibtex
+@misc{https://doi.org/10.48550/arxiv.2204.12484,
+  doi = {10.48550/ARXIV.2204.12484},
+  url = {https://arxiv.org/abs/2204.12484},
+  author = {Xu, Yufei and Zhang, Jing and Zhang, Qiming and Tao, Dacheng},
+  keywords = {Computer Vision and Pattern Recognition (cs.CV), FOS: Computer and information sciences, FOS: Computer and information sciences},
+  title = {ViTPose: Simple Vision Transformer Baselines for Human Pose Estimation},
+  publisher = {arXiv},
+  year = {2022},
+  copyright = {arXiv.org perpetual, non-exclusive license}
+}
+```
+
+</details>
+
+<!-- [DATASET] -->
+
+<details>
+<summary align="right"><a href="https://link.springer.com/chapter/10.1007/978-3-030-58545-7_12">COCO-WholeBody (ECCV'2020)</a></summary>
+
+```bibtex
+@inproceedings{jin2020whole,
+  title={Whole-Body Human Pose Estimation in the Wild},
+  author={Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping},
+  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
+  year={2020}
+}
+```
+
+</details>
+
+Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset
+
+> With classic decoder
+
+| Arch                                                                                                        | Input Size |  AP   | AP<sup>50</sup> | AP<sup>75</sup> |  AR   | AR<sup>50</sup> |   ckpt   |
+| :---------------------------------------------------------------------------------------------------------- | :--------: | :---: | :-------------: | :-------------: | :---: | :-------------: | :------: |
+| [ViTPose-S](/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-small_8xb64-210e_coco-256x192.py)  |  256x192   | 0.739 |      0.903      |      0.816      | 0.792 |      0.942      | \[ckpt\] |
+| [ViTPose-B](/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-base_8xb64-210e_coco-256x192.py)   |  256x192   | 0.757 |      0.905      |      0.829      | 0.810 |      0.946      | \[ckpt\] |
+| [ViTPose-L](/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-large_8xb64-210e_coco-256x192.py)  |  256x192   | 0.782 |      0.914      |      0.850      | 0.834 |      0.952      | \[ckpt\] |
+| [ViTPose-H](/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-huge_8xb64-210e_coco-256x192.py)   |  256x192   | 0.788 |      0.917      |      0.855      | 0.839 |      0.954      | \[ckpt\] |
+| [ViTPose-H\*](/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-huge_8xb64-210e_coco-256x192.py) |  256x192   | 0.790 |     0.0.916     |      0.857      | 0.840 |      0.953      | \[ckpt\] |
+
+*Models with * are converted from the [official repo](https://github.com/ViTAE-Transformer/ViTPose).  The config files of these models are only for validation.*
+
+> With simple decoder
+
+| Arch                                                                                                              | Input Size |  AP   | AP<sup>50</sup> | AP<sup>75</sup> |  AR   | AR<sup>50</sup> |    ckpt    |
+| :---------------------------------------------------------------------------------------------------------------- | :--------: | :---: | :-------------: | :-------------: | :---: | :-------------: | :--------: |
+| [ViTPose-S](/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-small-simple_8xb64-210e_coco-256x192.py) |  256x192   | 0.736 |      0.900      |      0.811      | 0.790 |      0.940      |  \[ckpt\]  |
+| [ViTPose-B](/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-base-simple_8xb64-210e_coco-256x192.py)  |  256x192   | 0.756 |      0.906      |      0.826      | 0.809 |      0.946      |  \[ckpt\]  |
+| [ViTPose-L](/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-large-simple_8xb64-210e_coco-256x192.py) |  256x192   | 0.781 |      0.914      |      0.853      | 0.833 |      0.952      | [ckpt](<>) |
+| [ViTPose-H](/configs/body_2d_keypoint/topdown_heatmap/coco/td-hm_ViTPose-huge-simple_8xb64-210e_coco-256x192.py)  |  256x192   | 0.789 |      0.916      |      0.856      | 0.839 |      0.953      |  \[ckpt\]  |
diff --git a/docs/src/papers/algorithms/vitpose.md b/docs/src/papers/algorithms/vitpose.md
new file mode 100644
index 0000000000..99fc2650f0
--- /dev/null
+++ b/docs/src/papers/algorithms/vitpose.md
@@ -0,0 +1,30 @@
+# Deep high-resolution representation learning for human pose estimation
+
+<!-- [ALGORITHM] -->
+
+<details>
+
+<summary  align="right"><a  href="https://arxiv.org/abs/2204.12484">ViTPose</a></summary>
+
+```bibtex
+
+@misc{https://doi.org/10.48550/arxiv.2204.12484,
+  doi = {10.48550/ARXIV.2204.12484},
+  url = {https://arxiv.org/abs/2204.12484},
+  author = {Xu, Yufei and Zhang, Jing and Zhang, Qiming and Tao, Dacheng},
+  keywords = {Computer Vision and Pattern Recognition (cs.CV), FOS: Computer and information sciences, FOS: Computer and information sciences},
+  title = {ViTPose: Simple Vision Transformer Baselines for Human Pose Estimation},
+  publisher = {arXiv},
+  year = {2022},
+  copyright = {arXiv.org perpetual, non-exclusive license}
+}
+
+```
+
+</details>
+
+## Abstract
+
+<!-- [ABSTRACT] -->
+
+Although no specific domain knowledge is considered in the design, plain vision transformers have shown excellent performance in visual recognition tasks. However, little effort has been made to reveal the potential of such simple structures for pose estimation tasks. In this paper, we show the surprisingly good capabilities of plain vision transformers for pose estimation from various aspects, namely simplicity in model structure, scalability in model size, flexibility in training paradigm, and transferability of knowledge between models, through a simple baseline model called ViTPose. Specifically, ViTPose employs plain and non-hierarchical vision transformers as backbones to extract features for a given person instance and a lightweight decoder for pose estimation. It can be scaled up from 100M to 1B parameters by taking the advantages of the scalable model capacity and high parallelism of transformers, setting a new Pareto front between throughput and performance. Besides, ViTPose is very flexible regarding the attention type, input resolution, pre-training and finetuning strategy, as well as dealing with multiple pose tasks. We also empirically demonstrate that the knowledge of large ViTPose models can be easily transferred to small ones via a simple knowledge token. Experimental results show that our basic ViTPose model outperforms representative methods on the challenging MS COCO Keypoint Detection benchmark, while the largest model sets a new state-of-the-art, i.e., 80.9 AP on the MS COCO test-dev set.
diff --git a/mmpose/engine/optim_wrappers/__init__.py b/mmpose/engine/optim_wrappers/__init__.py
new file mode 100644
index 0000000000..7c0b1f533a
--- /dev/null
+++ b/mmpose/engine/optim_wrappers/__init__.py
@@ -0,0 +1,4 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from .layer_decay_optim_wrapper import LayerDecayOptimWrapperConstructor
+
+__all__ = ['LayerDecayOptimWrapperConstructor']
diff --git a/mmpose/engine/optim_wrappers/layer_decay_optim_wrapper.py b/mmpose/engine/optim_wrappers/layer_decay_optim_wrapper.py
new file mode 100644
index 0000000000..6513e5593d
--- /dev/null
+++ b/mmpose/engine/optim_wrappers/layer_decay_optim_wrapper.py
@@ -0,0 +1,73 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from mmengine.dist.utils import get_dist_info
+from mmengine.optim import DefaultOptimWrapperConstructor
+from mmengine.registry import OPTIM_WRAPPER_CONSTRUCTORS
+
+
+def get_num_layer_for_vit(var_name, num_max_layer):
+    if var_name in ('backbone.cls_token', 'backbone.mask_token',
+                    'backbone.pos_embed'):
+        return 0
+    elif var_name.startswith('backbone.patch_embed'):
+        return 0
+    elif var_name.startswith('backbone.layers'):
+        layer_id = int(var_name.split('.')[2])
+        return layer_id + 1
+    else:
+        return num_max_layer - 1
+
+
+@OPTIM_WRAPPER_CONSTRUCTORS.register_module(force=True)
+class LayerDecayOptimWrapperConstructor(DefaultOptimWrapperConstructor):
+
+    def __init__(self, optim_wrapper_cfg, paramwise_cfg=None):
+        super().__init__(optim_wrapper_cfg, paramwise_cfg=None)
+        self.layer_decay_rate = paramwise_cfg.get('layer_decay_rate', 0.5)
+
+        super().__init__(optim_wrapper_cfg, paramwise_cfg)
+
+    def add_params(self, params, module, prefix='', lr=None):
+        parameter_groups = {}
+        print(self.paramwise_cfg)
+        num_layers = self.paramwise_cfg.get('num_layers') + 2
+        layer_decay_rate = self.paramwise_cfg.get('layer_decay_rate')
+        weight_decay = self.base_wd
+
+        for name, param in module.named_parameters():
+            if not param.requires_grad:
+                continue  # frozen weights
+            if (len(param.shape) == 1 or name.endswith('.bias')
+                    or 'pos_embed' in name):
+                group_name = 'no_decay'
+                this_weight_decay = 0.
+            else:
+                group_name = 'decay'
+                this_weight_decay = weight_decay
+            layer_id = get_num_layer_for_vit(name, num_layers)
+            group_name = 'layer_%d_%s' % (layer_id, group_name)
+
+            if group_name not in parameter_groups:
+                scale = layer_decay_rate**(num_layers - layer_id - 1)
+
+                parameter_groups[group_name] = {
+                    'weight_decay': this_weight_decay,
+                    'params': [],
+                    'param_names': [],
+                    'lr_scale': scale,
+                    'group_name': group_name,
+                    'lr': scale * self.base_lr,
+                }
+
+            parameter_groups[group_name]['params'].append(param)
+            parameter_groups[group_name]['param_names'].append(name)
+        rank, _ = get_dist_info()
+        if rank == 0:
+            to_display = {}
+            for key in parameter_groups:
+                to_display[key] = {
+                    'param_names': parameter_groups[key]['param_names'],
+                    'lr_scale': parameter_groups[key]['lr_scale'],
+                    'lr': parameter_groups[key]['lr'],
+                    'weight_decay': parameter_groups[key]['weight_decay'],
+                }
+        params.extend(parameter_groups.values())
diff --git a/mmpose/models/heads/base_head.py b/mmpose/models/heads/base_head.py
index e34c9363a5..40da595051 100644
--- a/mmpose/models/heads/base_head.py
+++ b/mmpose/models/heads/base_head.py
@@ -9,6 +9,7 @@
 from mmengine.structures import InstanceData
 from torch import Tensor
 
+from mmpose.models.utils.ops import resize
 from mmpose.utils.tensor_utils import to_numpy
 from mmpose.utils.typing import (Features, InstanceList, OptConfigType,
                                  OptSampleList, Predictions)
@@ -88,6 +89,12 @@ def _transform_inputs(
         elif self.input_transform == 'select':
             if isinstance(self.input_index, int):
                 inputs = feats[self.input_index]
+                if hasattr(self, 'upsample') and self.upsample > 0:
+                    inputs = resize(
+                        input=F.relu(inputs),
+                        scale_factor=self.upsample,
+                        mode='bilinear',
+                        align_corners=self.align_corners)
             else:
                 inputs = tuple(feats[i] for i in self.input_index)
         else:
diff --git a/mmpose/models/heads/heatmap_heads/heatmap_head.py b/mmpose/models/heads/heatmap_heads/heatmap_head.py
index 9a06cef16d..02ca7a893a 100644
--- a/mmpose/models/heads/heatmap_heads/heatmap_head.py
+++ b/mmpose/models/heads/heatmap_heads/heatmap_head.py
@@ -64,6 +64,8 @@ class HeatmapHead(BaseHead):
             keypoint coordinates from the network output. Defaults to ``None``
         init_cfg (Config, optional): Config to control the initialization. See
             :attr:`default_init_cfg` for default settings
+        extra (dict, optional): Extra configurations.
+            Defaults to ``None``
 
     .. _`Simple Baselines`: https://arxiv.org/abs/1804.06208
     """
@@ -84,7 +86,8 @@ def __init__(self,
                  loss: ConfigType = dict(
                      type='KeypointMSELoss', use_target_weight=True),
                  decoder: OptConfigType = None,
-                 init_cfg: OptConfigType = None):
+                 init_cfg: OptConfigType = None,
+                 extra=None):
 
         if init_cfg is None:
             init_cfg = self.default_init_cfg
@@ -101,6 +104,21 @@ def __init__(self,
             self.decoder = KEYPOINT_CODECS.build(decoder)
         else:
             self.decoder = None
+        self.upsample = 0
+
+        if extra is not None and not isinstance(extra, dict):
+            raise TypeError('extra should be dict or None.')
+
+        kernel_size = 1
+        padding = 0
+        if extra is not None:
+            if 'upsample' in extra:
+                self.upsample = extra['upsample']
+            if 'final_conv_kernel' in extra:
+                assert extra['final_conv_kernel'] in [1, 3]
+                if extra['final_conv_kernel'] == 3:
+                    padding = 1
+                kernel_size = extra['final_conv_kernel']
 
         # Get model input channels according to feature
         in_channels = self._get_in_channels()
@@ -149,7 +167,8 @@ def __init__(self,
                 type='Conv2d',
                 in_channels=in_channels,
                 out_channels=out_channels,
-                kernel_size=1)
+                padding=padding,
+                kernel_size=kernel_size)
             self.final_layer = build_conv_layer(cfg)
         else:
             self.final_layer = nn.Identity()