add new model

configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/mobilenetv2_coco.md

+<!-- [ALGORITHM] -->
+
+<details>
+<summary align="right"><a href="https://arxiv.org/abs/1611.05424">Associative Embedding (NIPS'2017)</a></summary>
+
+```bibtex
+@inproceedings{newell2017associative,
+  title={Associative embedding: End-to-end learning for joint detection and grouping},
+  author={Newell, Alejandro and Huang, Zhiao and Deng, Jia},
+  booktitle={Advances in neural information processing systems},
+  pages={2277--2287},
+  year={2017}
+}
+```
+
+</details>
+
+<!-- [BACKBONE] -->
+
+<details>
+<summary align="right"><a href="http://openaccess.thecvf.com/content_cvpr_2018/html/Sandler_MobileNetV2_Inverted_Residuals_CVPR_2018_paper.html">MobilenetV2 (CVPR'2018)</a></summary>
+
+```bibtex
+@inproceedings{sandler2018mobilenetv2,
+  title={Mobilenetv2: Inverted residuals and linear bottlenecks},
+  author={Sandler, Mark and Howard, Andrew and Zhu, Menglong and Zhmoginov, Andrey and Chen, Liang-Chieh},
+  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
+  pages={4510--4520},
+  year={2018}
+}
+```
+
+</details>
+
+<!-- [DATASET] -->
+
+<details>
+<summary align="right"><a href="https://link.springer.com/chapter/10.1007/978-3-319-10602-1_48">COCO (ECCV'2014)</a></summary>
+
+```bibtex
+@inproceedings{lin2014microsoft,
+  title={Microsoft coco: Common objects in context},
+  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
+  booktitle={European conference on computer vision},
+  pages={740--755},
+  year={2014},
+  organization={Springer}
+}
+```
+
+</details>
+
+Results on COCO val2017 without multi-scale test
+
+| Arch                                          | Input Size |  AP   | AP<sup>50</sup> | AP<sup>75</sup> |  AR   | AR<sup>50</sup> |                     ckpt                      |                      log                      |
+| :-------------------------------------------- | :--------: | :---: | :-------------: | :-------------: | :---: | :-------------: | :-------------------------------------------: | :-------------------------------------------: |
+| [pose_mobilenetv2](/configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/mobilenetv2_coco_512x512.py) |  512x512   | 0.380 |      0.671      |      0.368      | 0.473 |      0.741      | [ckpt](https://download.openmmlab.com/mmpose/bottom_up/mobilenetv2_coco_512x512-4d96e309_20200816.pth) | [log](https://download.openmmlab.com/mmpose/bottom_up/mobilenetv2_coco_512x512_20200816.log.json) |
+
+Results on COCO val2017 with multi-scale test. 3 default scales (\[2, 1, 0.5\]) are used
+
+| Arch                                          | Input Size |  AP   | AP<sup>50</sup> | AP<sup>75</sup> |  AR   | AR<sup>50</sup> |                     ckpt                      |                      log                      |
+| :-------------------------------------------- | :--------: | :---: | :-------------: | :-------------: | :---: | :-------------: | :-------------------------------------------: | :-------------------------------------------: |
+| [pose_mobilenetv2](/configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/mobilenetv2_coco_512x512.py) |  512x512   | 0.442 |      0.696      |      0.422      | 0.517 |      0.766      | [ckpt](https://download.openmmlab.com/mmpose/bottom_up/mobilenetv2_coco_512x512-4d96e309_20200816.pth) | [log](https://download.openmmlab.com/mmpose/bottom_up/mobilenetv2_coco_512x512_20200816.log.json) |

configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/mobilenetv2_coco.yml

+Collections:
+- Name: MobilenetV2
+  Paper:
+    Title: 'Mobilenetv2: Inverted residuals and linear bottlenecks'
+    URL: http://openaccess.thecvf.com/content_cvpr_2018/html/Sandler_MobileNetV2_Inverted_Residuals_CVPR_2018_paper.html
+  README: https://github.com/open-mmlab/mmpose/blob/master/docs/en/papers/backbones/mobilenetv2.md
+Models:
+- Config: configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/mobilenetv2_coco_512x512.py
+  In Collection: MobilenetV2
+  Metadata:
+    Architecture: &id001
+    - Associative Embedding
+    - MobilenetV2
+    Training Data: COCO
+  Name: associative_embedding_mobilenetv2_coco_512x512
+  Results:
+  - Dataset: COCO
+    Metrics:
+      AP: 0.38
+      AP@0.5: 0.671
+      AP@0.75: 0.368
+      AR: 0.473
+      AR@0.5: 0.741
+    Task: Body 2D Keypoint
+  Weights: https://download.openmmlab.com/mmpose/bottom_up/mobilenetv2_coco_512x512-4d96e309_20200816.pth
+- Config: configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/mobilenetv2_coco_512x512.py
+  In Collection: MobilenetV2
+  Metadata:
+    Architecture: *id001
+    Training Data: COCO
+  Name: associative_embedding_mobilenetv2_coco_512x512
+  Results:
+  - Dataset: COCO
+    Metrics:
+      AP: 0.442
+      AP@0.5: 0.696
+      AP@0.75: 0.422
+      AR: 0.517
+      AR@0.5: 0.766
+    Task: Body 2D Keypoint
+  Weights: https://download.openmmlab.com/mmpose/bottom_up/mobilenetv2_coco_512x512-4d96e309_20200816.pth

configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/mobilenetv2_coco_512x512.py

+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/coco.py'
+]
+checkpoint_config = dict(interval=50)
+evaluation = dict(interval=50, metric='mAP', save_best='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 = 300
+channel_cfg = dict(
+    dataset_joints=17,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
+    ])
+
+data_cfg = dict(
+    image_size=512,
+    base_size=256,
+    base_sigma=2,
+    heatmap_size=[128],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    num_scales=1,
+    scale_aware_sigma=False,
+)
+
+# model settings
+model = dict(
+    type='AssociativeEmbedding',
+    pretrained='mmcls://mobilenet_v2',
+    backbone=dict(type='MobileNetV2', widen_factor=1., out_indices=(7, )),
+    keypoint_head=dict(
+        type='AESimpleHead',
+        in_channels=1280,
+        num_joints=17,
+        tag_per_joint=True,
+        with_ae_loss=[True],
+        loss_keypoint=dict(
+            type='MultiLossFactory',
+            num_joints=17,
+            num_stages=1,
+            ae_loss_type='exp',
+            with_ae_loss=[True],
+            push_loss_factor=[0.001],
+            pull_loss_factor=[0.001],
+            with_heatmaps_loss=[True],
+            heatmaps_loss_factor=[1.0])),
+    train_cfg=dict(),
+    test_cfg=dict(
+        num_joints=channel_cfg['dataset_joints'],
+        max_num_people=30,
+        scale_factor=[1],
+        with_heatmaps=[True],
+        with_ae=[True],
+        project2image=True,
+        align_corners=False,
+        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))
+
+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/coco'
+data = dict(
+    workers_per_gpu=1,
+    train_dataloader=dict(samples_per_gpu=24),
+    val_dataloader=dict(samples_per_gpu=1),
+    test_dataloader=dict(samples_per_gpu=1),
+    train=dict(
+        type='BottomUpCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
+        img_prefix=f'{data_root}/train2017/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='BottomUpCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='BottomUpCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)

configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/res101_coco_512x512.py

+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/coco.py'
+]
+checkpoint_config = dict(interval=50)
+evaluation = dict(interval=50, metric='mAP', save_best='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 = 300
+channel_cfg = dict(
+    dataset_joints=17,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
+    ])
+
+data_cfg = dict(
+    image_size=512,
+    base_size=256,
+    base_sigma=2,
+    heatmap_size=[128],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    num_scales=1,
+    scale_aware_sigma=False,
+)
+
+# model settings
+model = dict(
+    type='AssociativeEmbedding',
+    pretrained='torchvision://resnet101',
+    backbone=dict(type='ResNet', depth=101),
+    keypoint_head=dict(
+        type='AESimpleHead',
+        in_channels=2048,
+        num_joints=17,
+        tag_per_joint=True,
+        with_ae_loss=[True],
+        loss_keypoint=dict(
+            type='MultiLossFactory',
+            num_joints=17,
+            num_stages=1,
+            ae_loss_type='exp',
+            with_ae_loss=[True],
+            push_loss_factor=[0.001],
+            pull_loss_factor=[0.001],
+            with_heatmaps_loss=[True],
+            heatmaps_loss_factor=[1.0])),
+    train_cfg=dict(),
+    test_cfg=dict(
+        num_joints=channel_cfg['dataset_joints'],
+        max_num_people=30,
+        scale_factor=[1],
+        with_heatmaps=[True],
+        with_ae=[True],
+        project2image=True,
+        align_corners=False,
+        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))
+
+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/coco'
+data = dict(
+    workers_per_gpu=2,
+    train_dataloader=dict(samples_per_gpu=16),
+    val_dataloader=dict(samples_per_gpu=1),
+    test_dataloader=dict(samples_per_gpu=1),
+    train=dict(
+        type='BottomUpCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
+        img_prefix=f'{data_root}/train2017/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='BottomUpCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='BottomUpCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)

configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/res101_coco_640x640.py

+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/coco.py'
+]
+checkpoint_config = dict(interval=50)
+evaluation = dict(interval=50, metric='mAP', save_best='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 = 300
+channel_cfg = dict(
+    dataset_joints=17,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
+    ])
+
+data_cfg = dict(
+    image_size=640,
+    base_size=320,
+    base_sigma=2,
+    heatmap_size=[160],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    num_scales=1,
+    scale_aware_sigma=False,
+)
+
+# model settings
+model = dict(
+    type='AssociativeEmbedding',
+    pretrained='torchvision://resnet101',
+    backbone=dict(type='ResNet', depth=101),
+    keypoint_head=dict(
+        type='AESimpleHead',
+        in_channels=2048,
+        num_joints=17,
+        tag_per_joint=True,
+        with_ae_loss=[True],
+        loss_keypoint=dict(
+            type='MultiLossFactory',
+            num_joints=17,
+            num_stages=1,
+            ae_loss_type='exp',
+            with_ae_loss=[True],
+            push_loss_factor=[0.001],
+            pull_loss_factor=[0.001],
+            with_heatmaps_loss=[True],
+            heatmaps_loss_factor=[1.0])),
+    train_cfg=dict(),
+    test_cfg=dict(
+        num_joints=channel_cfg['dataset_joints'],
+        max_num_people=30,
+        scale_factor=[1],
+        with_heatmaps=[True],
+        with_ae=[True],
+        project2image=True,
+        align_corners=False,
+        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))
+
+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/coco'
+data = dict(
+    workers_per_gpu=2,
+    train_dataloader=dict(samples_per_gpu=16),
+    val_dataloader=dict(samples_per_gpu=1),
+    test_dataloader=dict(samples_per_gpu=1),
+    train=dict(
+        type='BottomUpCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
+        img_prefix=f'{data_root}/train2017/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='BottomUpCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='BottomUpCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)

configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/res152_coco_512x512.py

+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/coco.py'
+]
+checkpoint_config = dict(interval=50)
+evaluation = dict(interval=50, metric='mAP', save_best='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 = 300
+channel_cfg = dict(
+    dataset_joints=17,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
+    ])
+
+data_cfg = dict(
+    image_size=512,
+    base_size=256,
+    base_sigma=2,
+    heatmap_size=[128],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    num_scales=1,
+    scale_aware_sigma=False,
+)
+
+# model settings
+model = dict(
+    type='AssociativeEmbedding',
+    pretrained='torchvision://resnet152',
+    backbone=dict(type='ResNet', depth=152),
+    keypoint_head=dict(
+        type='AESimpleHead',
+        in_channels=2048,
+        num_joints=17,
+        tag_per_joint=True,
+        with_ae_loss=[True],
+        loss_keypoint=dict(
+            type='MultiLossFactory',
+            num_joints=17,
+            num_stages=1,
+            ae_loss_type='exp',
+            with_ae_loss=[True],
+            push_loss_factor=[0.001],
+            pull_loss_factor=[0.001],
+            with_heatmaps_loss=[True],
+            heatmaps_loss_factor=[1.0])),
+    train_cfg=dict(),
+    test_cfg=dict(
+        num_joints=channel_cfg['dataset_joints'],
+        max_num_people=30,
+        scale_factor=[1],
+        with_heatmaps=[True],
+        with_ae=[True],
+        project2image=True,
+        align_corners=False,
+        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))
+
+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/coco'
+data = dict(
+    workers_per_gpu=2,
+    train_dataloader=dict(samples_per_gpu=16),
+    val_dataloader=dict(samples_per_gpu=1),
+    test_dataloader=dict(samples_per_gpu=1),
+    train=dict(
+        type='BottomUpCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
+        img_prefix=f'{data_root}/train2017/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='BottomUpCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='BottomUpCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)

configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/res152_coco_640x640.py

+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/coco.py'
+]
+checkpoint_config = dict(interval=50)
+evaluation = dict(interval=50, metric='mAP', save_best='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 = 300
+channel_cfg = dict(
+    dataset_joints=17,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
+    ])
+
+data_cfg = dict(
+    image_size=640,
+    base_size=320,
+    base_sigma=2,
+    heatmap_size=[160],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    num_scales=1,
+    scale_aware_sigma=False,
+)
+
+# model settings
+model = dict(
+    type='AssociativeEmbedding',
+    pretrained='torchvision://resnet152',
+    backbone=dict(type='ResNet', depth=152),
+    keypoint_head=dict(
+        type='AESimpleHead',
+        in_channels=2048,
+        num_joints=17,
+        tag_per_joint=True,
+        with_ae_loss=[True],
+        loss_keypoint=dict(
+            type='MultiLossFactory',
+            num_joints=17,
+            num_stages=1,
+            ae_loss_type='exp',
+            with_ae_loss=[True],
+            push_loss_factor=[0.001],
+            pull_loss_factor=[0.001],
+            with_heatmaps_loss=[True],
+            heatmaps_loss_factor=[1.0])),
+    train_cfg=dict(),
+    test_cfg=dict(
+        num_joints=channel_cfg['dataset_joints'],
+        max_num_people=30,
+        scale_factor=[1],
+        with_heatmaps=[True],
+        with_ae=[True],
+        project2image=True,
+        align_corners=False,
+        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))
+
+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/coco'
+data = dict(
+    workers_per_gpu=2,
+    train_dataloader=dict(samples_per_gpu=16),
+    val_dataloader=dict(samples_per_gpu=1),
+    test_dataloader=dict(samples_per_gpu=1),
+    train=dict(
+        type='BottomUpCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
+        img_prefix=f'{data_root}/train2017/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='BottomUpCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='BottomUpCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)

configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/res50_coco_512x512.py

+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/coco.py'
+]
+checkpoint_config = dict(interval=50)
+evaluation = dict(interval=50, metric='mAP', save_best='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 = 300
+channel_cfg = dict(
+    dataset_joints=17,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
+    ])
+
+data_cfg = dict(
+    image_size=512,
+    base_size=256,
+    base_sigma=2,
+    heatmap_size=[128],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    num_scales=1,
+    scale_aware_sigma=False,
+)
+
+# model settings
+model = dict(
+    type='AssociativeEmbedding',
+    pretrained='torchvision://resnet50',
+    backbone=dict(type='ResNet', depth=50),
+    keypoint_head=dict(
+        type='AESimpleHead',
+        in_channels=2048,
+        num_joints=17,
+        tag_per_joint=True,
+        with_ae_loss=[True],
+        loss_keypoint=dict(
+            type='MultiLossFactory',
+            num_joints=17,
+            num_stages=1,
+            ae_loss_type='exp',
+            with_ae_loss=[True],
+            push_loss_factor=[0.001],
+            pull_loss_factor=[0.001],
+            with_heatmaps_loss=[True],
+            heatmaps_loss_factor=[1.0],
+        )),
+    train_cfg=dict(),
+    test_cfg=dict(
+        num_joints=channel_cfg['dataset_joints'],
+        max_num_people=30,
+        scale_factor=[1],
+        with_heatmaps=[True],
+        with_ae=[True],
+        project2image=True,
+        align_corners=False,
+        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))
+
+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/coco'
+data = dict(
+    workers_per_gpu=1,
+    train_dataloader=dict(samples_per_gpu=24),
+    val_dataloader=dict(samples_per_gpu=1),
+    test_dataloader=dict(samples_per_gpu=1),
+    train=dict(
+        type='BottomUpCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
+        img_prefix=f'{data_root}/train2017/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='BottomUpCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='BottomUpCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)

configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/res50_coco_640x640.py

+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/coco.py'
+]
+checkpoint_config = dict(interval=50)
+evaluation = dict(interval=50, metric='mAP', save_best='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 = 300
+channel_cfg = dict(
+    dataset_joints=17,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
+    ])
+
+data_cfg = dict(
+    image_size=640,
+    base_size=320,
+    base_sigma=2,
+    heatmap_size=[160],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    num_scales=1,
+    scale_aware_sigma=False,
+)
+
+# model settings
+model = dict(
+    type='AssociativeEmbedding',
+    pretrained='torchvision://resnet50',
+    backbone=dict(type='ResNet', depth=50),
+    keypoint_head=dict(
+        type='AESimpleHead',
+        in_channels=2048,
+        num_joints=17,
+        tag_per_joint=True,
+        with_ae_loss=[True],
+        loss_keypoint=dict(
+            type='MultiLossFactory',
+            num_joints=17,
+            num_stages=1,
+            ae_loss_type='exp',
+            with_ae_loss=[True],
+            push_loss_factor=[0.001],
+            pull_loss_factor=[0.001],
+            with_heatmaps_loss=[True],
+            heatmaps_loss_factor=[1.0])),
+    train_cfg=dict(),
+    test_cfg=dict(
+        num_joints=channel_cfg['dataset_joints'],
+        max_num_people=30,
+        scale_factor=[1],
+        with_heatmaps=[True],
+        with_ae=[True],
+        project2image=True,
+        align_corners=False,
+        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))
+
+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/coco'
+data = dict(
+    workers_per_gpu=1,
+    train_dataloader=dict(samples_per_gpu=24),
+    val_dataloader=dict(samples_per_gpu=1),
+    test_dataloader=dict(samples_per_gpu=1),
+    train=dict(
+        type='BottomUpCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
+        img_prefix=f'{data_root}/train2017/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='BottomUpCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='BottomUpCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)

configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/resnet_coco.md

+<!-- [ALGORITHM] -->
+
+<details>
+<summary align="right"><a href="https://arxiv.org/abs/1611.05424">Associative Embedding (NIPS'2017)</a></summary>
+
+```bibtex
+@inproceedings{newell2017associative,
+  title={Associative embedding: End-to-end learning for joint detection and grouping},
+  author={Newell, Alejandro and Huang, Zhiao and Deng, Jia},
+  booktitle={Advances in neural information processing systems},
+  pages={2277--2287},
+  year={2017}
+}
+```
+
+</details>
+
+<!-- [BACKBONE] -->
+
+<details>
+<summary align="right"><a href="http://openaccess.thecvf.com/content_cvpr_2016/html/He_Deep_Residual_Learning_CVPR_2016_paper.html">ResNet (CVPR'2016)</a></summary>
+
+```bibtex
+@inproceedings{he2016deep,
+  title={Deep residual learning for image recognition},
+  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
+  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
+  pages={770--778},
+  year={2016}
+}
+```
+
+</details>
+
+<!-- [DATASET] -->
+
+<details>
+<summary align="right"><a href="https://link.springer.com/chapter/10.1007/978-3-319-10602-1_48">COCO (ECCV'2014)</a></summary>
+
+```bibtex
+@inproceedings{lin2014microsoft,
+  title={Microsoft coco: Common objects in context},
+  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
+  booktitle={European conference on computer vision},
+  pages={740--755},
+  year={2014},
+  organization={Springer}
+}
+```
+
+</details>
+
+Results on COCO val2017 without multi-scale test
+
+| Arch                                          | Input Size |  AP   | AP<sup>50</sup> | AP<sup>75</sup> |  AR   | AR<sup>50</sup> |                     ckpt                      |                      log                      |
+| :-------------------------------------------- | :--------: | :---: | :-------------: | :-------------: | :---: | :-------------: | :-------------------------------------------: | :-------------------------------------------: |
+| [pose_resnet_50](/configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/res50_coco_512x512.py) |  512x512   | 0.466 |      0.742      |      0.479      | 0.552 |      0.797      | [ckpt](https://download.openmmlab.com/mmpose/bottom_up/res50_coco_512x512-5521bead_20200816.pth) | [log](https://download.openmmlab.com/mmpose/bottom_up/res50_coco_512x512_20200816.log.json) |
+| [pose_resnet_50](/configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/res50_coco_640x640.py) |  640x640   | 0.479 |      0.757      |      0.487      | 0.566 |      0.810      | [ckpt](https://download.openmmlab.com/mmpose/bottom_up/res50_coco_640x640-2046f9cb_20200822.pth) | [log](https://download.openmmlab.com/mmpose/bottom_up/res50_coco_640x640_20200822.log.json) |
+| [pose_resnet_101](/configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/res101_coco_512x512.py) |  512x512   | 0.554 |      0.807      |      0.599      | 0.622 |      0.841      | [ckpt](https://download.openmmlab.com/mmpose/bottom_up/res101_coco_512x512-e0c95157_20200816.pth) | [log](https://download.openmmlab.com/mmpose/bottom_up/res101_coco_512x512_20200816.log.json) |
+| [pose_resnet_152](/configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/res152_coco_512x512.py) |  512x512   | 0.595 |      0.829      |      0.648      | 0.651 |      0.856      | [ckpt](https://download.openmmlab.com/mmpose/bottom_up/res152_coco_512x512-364eb38d_20200822.pth) | [log](https://download.openmmlab.com/mmpose/bottom_up/res152_coco_512x512_20200822.log.json) |
+
+Results on COCO val2017 with multi-scale test. 3 default scales (\[2, 1, 0.5\]) are used
+
+| Arch                                          | Input Size |  AP   | AP<sup>50</sup> | AP<sup>75</sup> |  AR   | AR<sup>50</sup> |                     ckpt                      |                      log                      |
+| :-------------------------------------------- | :--------: | :---: | :-------------: | :-------------: | :---: | :-------------: | :-------------------------------------------: | :-------------------------------------------: |
+| [pose_resnet_50](/configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/res50_coco_512x512.py) |  512x512   | 0.503 |      0.765      |      0.521      | 0.591 |      0.821      | [ckpt](https://download.openmmlab.com/mmpose/bottom_up/res50_coco_512x512-5521bead_20200816.pth) | [log](https://download.openmmlab.com/mmpose/bottom_up/res50_coco_512x512_20200816.log.json) |
+| [pose_resnet_50](/configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/res50_coco_640x640.py) |  640x640   | 0.525 |      0.784      |      0.542      | 0.610 |      0.832      | [ckpt](https://download.openmmlab.com/mmpose/bottom_up/res50_coco_640x640-2046f9cb_20200822.pth) | [log](https://download.openmmlab.com/mmpose/bottom_up/res50_coco_640x640_20200822.log.json) |
+| [pose_resnet_101](/configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/res101_coco_512x512.py) |  512x512   | 0.603 |      0.831      |      0.641      | 0.668 |      0.870      | [ckpt](https://download.openmmlab.com/mmpose/bottom_up/res101_coco_512x512-e0c95157_20200816.pth) | [log](https://download.openmmlab.com/mmpose/bottom_up/res101_coco_512x512_20200816.log.json) |
+| [pose_resnet_152](/configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/res152_coco_512x512.py) |  512x512   | 0.660 |      0.860      |      0.713      | 0.709 |      0.889      | [ckpt](https://download.openmmlab.com/mmpose/bottom_up/res152_coco_512x512-364eb38d_20200822.pth) | [log](https://download.openmmlab.com/mmpose/bottom_up/res152_coco_512x512_20200822.log.json) |

configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/resnet_coco.yml

+Collections:
+- Name: Associative Embedding
+  Paper:
+    Title: 'Associative embedding: End-to-end learning for joint detection and grouping'
+    URL: https://arxiv.org/abs/1611.05424
+  README: https://github.com/open-mmlab/mmpose/blob/master/docs/en/papers/algorithms/associative_embedding.md
+Models:
+- Config: configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/res50_coco_512x512.py
+  In Collection: Associative Embedding
+  Metadata:
+    Architecture: &id001
+    - Associative Embedding
+    - ResNet
+    Training Data: COCO
+  Name: associative_embedding_res50_coco_512x512
+  Results:
+  - Dataset: COCO
+    Metrics:
+      AP: 0.466
+      AP@0.5: 0.742
+      AP@0.75: 0.479
+      AR: 0.552
+      AR@0.5: 0.797
+    Task: Body 2D Keypoint
+  Weights: https://download.openmmlab.com/mmpose/bottom_up/res50_coco_512x512-5521bead_20200816.pth
+- Config: configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/res50_coco_640x640.py
+  In Collection: Associative Embedding
+  Metadata:
+    Architecture: *id001
+    Training Data: COCO
+  Name: associative_embedding_res50_coco_640x640
+  Results:
+  - Dataset: COCO
+    Metrics:
+      AP: 0.479
+      AP@0.5: 0.757
+      AP@0.75: 0.487
+      AR: 0.566
+      AR@0.5: 0.81
+    Task: Body 2D Keypoint
+  Weights: https://download.openmmlab.com/mmpose/bottom_up/res50_coco_640x640-2046f9cb_20200822.pth
+- Config: configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/res101_coco_512x512.py
+  In Collection: Associative Embedding
+  Metadata:
+    Architecture: *id001
+    Training Data: COCO
+  Name: associative_embedding_res101_coco_512x512
+  Results:
+  - Dataset: COCO
+    Metrics:
+      AP: 0.554
+      AP@0.5: 0.807
+      AP@0.75: 0.599
+      AR: 0.622
+      AR@0.5: 0.841
+    Task: Body 2D Keypoint
+  Weights: https://download.openmmlab.com/mmpose/bottom_up/res101_coco_512x512-e0c95157_20200816.pth
+- Config: configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/res152_coco_512x512.py
+  In Collection: Associative Embedding
+  Metadata:
+    Architecture: *id001
+    Training Data: COCO
+  Name: associative_embedding_res152_coco_512x512
+  Results:
+  - Dataset: COCO
+    Metrics:
+      AP: 0.595
+      AP@0.5: 0.829
+      AP@0.75: 0.648
+      AR: 0.651
+      AR@0.5: 0.856
+    Task: Body 2D Keypoint
+  Weights: https://download.openmmlab.com/mmpose/bottom_up/res152_coco_512x512-364eb38d_20200822.pth
+- Config: configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/res50_coco_512x512.py
+  In Collection: Associative Embedding
+  Metadata:
+    Architecture: *id001
+    Training Data: COCO
+  Name: associative_embedding_res50_coco_512x512
+  Results:
+  - Dataset: COCO
+    Metrics:
+      AP: 0.503
+      AP@0.5: 0.765
+      AP@0.75: 0.521
+      AR: 0.591
+      AR@0.5: 0.821
+    Task: Body 2D Keypoint
+  Weights: https://download.openmmlab.com/mmpose/bottom_up/res50_coco_512x512-5521bead_20200816.pth
+- Config: configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/res50_coco_640x640.py
+  In Collection: Associative Embedding
+  Metadata:
+    Architecture: *id001
+    Training Data: COCO
+  Name: associative_embedding_res50_coco_640x640
+  Results:
+  - Dataset: COCO
+    Metrics:
+      AP: 0.525
+      AP@0.5: 0.784
+      AP@0.75: 0.542
+      AR: 0.61
+      AR@0.5: 0.832
+    Task: Body 2D Keypoint
+  Weights: https://download.openmmlab.com/mmpose/bottom_up/res50_coco_640x640-2046f9cb_20200822.pth
+- Config: configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/res101_coco_512x512.py
+  In Collection: Associative Embedding
+  Metadata:
+    Architecture: *id001
+    Training Data: COCO
+  Name: associative_embedding_res101_coco_512x512
+  Results:
+  - Dataset: COCO
+    Metrics:
+      AP: 0.603
+      AP@0.5: 0.831
+      AP@0.75: 0.641
+      AR: 0.668
+      AR@0.5: 0.87
+    Task: Body 2D Keypoint
+  Weights: https://download.openmmlab.com/mmpose/bottom_up/res101_coco_512x512-e0c95157_20200816.pth
+- Config: configs/body/2d_kpt_sview_rgb_img/associative_embedding/coco/res152_coco_512x512.py
+  In Collection: Associative Embedding
+  Metadata:
+    Architecture: *id001
+    Training Data: COCO
+  Name: associative_embedding_res152_coco_512x512
+  Results:
+  - Dataset: COCO
+    Metrics:
+      AP: 0.66
+      AP@0.5: 0.86
+      AP@0.75: 0.713
+      AR: 0.709
+      AR@0.5: 0.889
+    Task: Body 2D Keypoint
+  Weights: https://download.openmmlab.com/mmpose/bottom_up/res152_coco_512x512-364eb38d_20200822.pth

configs/body/2d_kpt_sview_rgb_img/associative_embedding/crowdpose/higherhrnet_crowdpose.md

+<!-- [ALGORITHM] -->
+
+<details>
+<summary align="right"><a href="https://arxiv.org/abs/1611.05424">Associative Embedding (NIPS'2017)</a></summary>
+
+```bibtex
+@inproceedings{newell2017associative,
+  title={Associative embedding: End-to-end learning for joint detection and grouping},
+  author={Newell, Alejandro and Huang, Zhiao and Deng, Jia},
+  booktitle={Advances in neural information processing systems},
+  pages={2277--2287},
+  year={2017}
+}
+```
+
+</details>
+
+<!-- [ALGORITHM] -->
+
+<details>
+<summary align="right"><a href="http://openaccess.thecvf.com/content_CVPR_2020/html/Cheng_HigherHRNet_Scale-Aware_Representation_Learning_for_Bottom-Up_Human_Pose_Estimation_CVPR_2020_paper.html">HigherHRNet (CVPR'2020)</a></summary>
+
+```bibtex
+@inproceedings{cheng2020higherhrnet,
+  title={HigherHRNet: Scale-Aware Representation Learning for Bottom-Up Human Pose Estimation},
+  author={Cheng, Bowen and Xiao, Bin and Wang, Jingdong and Shi, Honghui and Huang, Thomas S and Zhang, Lei},
+  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
+  pages={5386--5395},
+  year={2020}
+}
+```
+
+</details>
+
+<!-- [DATASET] -->
+
+<details>
+<summary align="right"><a href="http://openaccess.thecvf.com/content_CVPR_2019/html/Li_CrowdPose_Efficient_Crowded_Scenes_Pose_Estimation_and_a_New_Benchmark_CVPR_2019_paper.html">CrowdPose (CVPR'2019)</a></summary>
+
+```bibtex
+@article{li2018crowdpose,
+  title={CrowdPose: Efficient Crowded Scenes Pose Estimation and A New Benchmark},
+  author={Li, Jiefeng and Wang, Can and Zhu, Hao and Mao, Yihuan and Fang, Hao-Shu and Lu, Cewu},
+  journal={arXiv preprint arXiv:1812.00324},
+  year={2018}
+}
+```
+
+</details>
+
+Results on CrowdPose test without multi-scale test
+
+| Arch                                           | Input Size |  AP   | AP<sup>50</sup> | AP<sup>75</sup> | AP (E) | AP (M) | AP (H) |                      ckpt                      |                      log                      |
+| :--------------------------------------------- | :--------: | :---: | :-------------: | :-------------: | :----: | :----: | :----: | :--------------------------------------------: | :-------------------------------------------: |
+| [HigherHRNet-w32](/configs/body/2d_kpt_sview_rgb_img/associative_embedding/crowdpose/higherhrnet_w32_crowdpose_512x512.py) |  512x512   | 0.655 |      0.859      |      0.705      | 0.728  | 0.660  | 0.577  | [ckpt](https://download.openmmlab.com/mmpose/bottom_up/higher_hrnet32_crowdpose_512x512-1aa4a132_20201017.pth) | [log](https://download.openmmlab.com/mmpose/bottom_up/higher_hrnet32_crowdpose_512x512_20201017.log.json) |
+
+Results on CrowdPose test with multi-scale test. 2 scales (\[2, 1\]) are used
+
+| Arch                                           | Input Size |  AP   | AP<sup>50</sup> | AP<sup>75</sup> | AP (E) | AP (M) | AP (H) |                      ckpt                      |                      log                      |
+| :--------------------------------------------- | :--------: | :---: | :-------------: | :-------------: | :----: | :----: | :----: | :--------------------------------------------: | :-------------------------------------------: |
+| [HigherHRNet-w32](/configs/body/2d_kpt_sview_rgb_img/associative_embedding/crowdpose/higherhrnet_w32_crowdpose_512x512.py) |  512x512   | 0.661 |      0.864      |      0.710      | 0.742  | 0.670  | 0.566  | [ckpt](https://download.openmmlab.com/mmpose/bottom_up/higher_hrnet32_crowdpose_512x512-1aa4a132_20201017.pth) | [log](https://download.openmmlab.com/mmpose/bottom_up/higher_hrnet32_crowdpose_512x512_20201017.log.json) |

configs/body/2d_kpt_sview_rgb_img/associative_embedding/crowdpose/higherhrnet_crowdpose.yml

+Collections:
+- Name: HigherHRNet
+  Paper:
+    Title: 'HigherHRNet: Scale-Aware Representation Learning for Bottom-Up Human Pose
+      Estimation'
+    URL: http://openaccess.thecvf.com/content_CVPR_2020/html/Cheng_HigherHRNet_Scale-Aware_Representation_Learning_for_Bottom-Up_Human_Pose_Estimation_CVPR_2020_paper.html
+  README: https://github.com/open-mmlab/mmpose/blob/master/docs/en/papers/backbones/higherhrnet.md
+Models:
+- Config: configs/body/2d_kpt_sview_rgb_img/associative_embedding/crowdpose/higherhrnet_w32_crowdpose_512x512.py
+  In Collection: HigherHRNet
+  Metadata:
+    Architecture: &id001
+    - Associative Embedding
+    - HigherHRNet
+    Training Data: CrowdPose
+  Name: associative_embedding_higherhrnet_w32_crowdpose_512x512
+  Results:
+  - Dataset: CrowdPose
+    Metrics:
+      AP: 0.655
+      AP (E): 0.728
+      AP (H): 0.577
+      AP (M): 0.66
+      AP@0.5: 0.859
+      AP@0.75: 0.705
+    Task: Body 2D Keypoint
+  Weights: https://download.openmmlab.com/mmpose/bottom_up/higher_hrnet32_crowdpose_512x512-1aa4a132_20201017.pth
+- Config: configs/body/2d_kpt_sview_rgb_img/associative_embedding/crowdpose/higherhrnet_w32_crowdpose_512x512.py
+  In Collection: HigherHRNet
+  Metadata:
+    Architecture: *id001
+    Training Data: CrowdPose
+  Name: associative_embedding_higherhrnet_w32_crowdpose_512x512
+  Results:
+  - Dataset: CrowdPose
+    Metrics:
+      AP: 0.661
+      AP (E): 0.742
+      AP (H): 0.566
+      AP (M): 0.67
+      AP@0.5: 0.864
+      AP@0.75: 0.71
+    Task: Body 2D Keypoint
+  Weights: https://download.openmmlab.com/mmpose/bottom_up/higher_hrnet32_crowdpose_512x512-1aa4a132_20201017.pth

configs/body/2d_kpt_sview_rgb_img/associative_embedding/crowdpose/higherhrnet_w32_crowdpose_512x512.py

+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/crowdpose.py'
+]
+checkpoint_config = dict(interval=50)
+evaluation = dict(interval=50, metric='mAP', save_best='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 = 300
+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='AssociativeEmbedding',
+    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='AEHigherResolutionHead',
+        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],
+        loss_keypoint=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_cfg=dict(),
+    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,
+        align_corners=False,
+        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))
+
+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/crowdpose'
+data = dict(
+    workers_per_gpu=2,
+    train_dataloader=dict(samples_per_gpu=24),
+    val_dataloader=dict(samples_per_gpu=1),
+    test_dataloader=dict(samples_per_gpu=1),
+    train=dict(
+        type='BottomUpCrowdPoseDataset',
+        ann_file=f'{data_root}/annotations/mmpose_crowdpose_trainval.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='BottomUpCrowdPoseDataset',
+        ann_file=f'{data_root}/annotations/mmpose_crowdpose_test.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='BottomUpCrowdPoseDataset',
+        ann_file=f'{data_root}/annotations/mmpose_crowdpose_test.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)

configs/body/2d_kpt_sview_rgb_img/associative_embedding/crowdpose/higherhrnet_w32_crowdpose_512x512_udp.py

+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/crowdpose.py'
+]
+checkpoint_config = dict(interval=50)
+evaluation = dict(interval=50, metric='mAP', save_best='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 = 300
+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='AssociativeEmbedding',
+    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='AEHigherResolutionHead',
+        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],
+        loss_keypoint=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_cfg=dict(),
+    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=False,
+        align_corners=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,
+        use_udp=True))
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='BottomUpRandomAffine',
+        rot_factor=30,
+        scale_factor=[0.75, 1.5],
+        scale_type='short',
+        trans_factor=40,
+        use_udp=True),
+    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,
+        use_udp=True,
+    ),
+    dict(
+        type='Collect',
+        keys=['img', 'joints', 'targets', 'masks'],
+        meta_keys=[]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='BottomUpGetImgSize', test_scale_factor=[1], use_udp=True),
+    dict(
+        type='BottomUpResizeAlign',
+        transforms=[
+            dict(type='ToTensor'),
+            dict(
+                type='NormalizeTensor',
+                mean=[0.485, 0.456, 0.406],
+                std=[0.229, 0.224, 0.225])
+        ],
+        use_udp=True),
+    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/crowdpose'
+data = dict(
+    workers_per_gpu=2,
+    train_dataloader=dict(samples_per_gpu=24),
+    val_dataloader=dict(samples_per_gpu=1),
+    test_dataloader=dict(samples_per_gpu=1),
+    train=dict(
+        type='BottomUpCrowdPoseDataset',
+        ann_file=f'{data_root}/annotations/mmpose_crowdpose_trainval.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='BottomUpCrowdPoseDataset',
+        ann_file=f'{data_root}/annotations/mmpose_crowdpose_test.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='BottomUpCrowdPoseDataset',
+        ann_file=f'{data_root}/annotations/mmpose_crowdpose_test.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)

configs/body/2d_kpt_sview_rgb_img/associative_embedding/crowdpose/higherhrnet_w32_crowdpose_640x640.py

+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/crowdpose.py'
+]
+checkpoint_config = dict(interval=50)
+evaluation = dict(interval=50, metric='mAP', save_best='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 = 300
+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=640,
+    base_size=320,
+    base_sigma=2,
+    heatmap_size=[160, 320],
+    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='AssociativeEmbedding',
+    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='AEHigherResolutionHead',
+        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],
+        loss_keypoint=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_cfg=dict(),
+    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,
+        align_corners=False,
+        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))
+
+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/crowdpose'
+data = dict(
+    workers_per_gpu=2,
+    train_dataloader=dict(samples_per_gpu=16),
+    val_dataloader=dict(samples_per_gpu=1),
+    test_dataloader=dict(samples_per_gpu=1),
+    train=dict(
+        type='BottomUpCrowdPoseDataset',
+        ann_file=f'{data_root}/annotations/mmpose_crowdpose_trainval.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='BottomUpCrowdPoseDataset',
+        ann_file=f'{data_root}/annotations/mmpose_crowdpose_test.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='BottomUpCrowdPoseDataset',
+        ann_file=f'{data_root}/annotations/mmpose_crowdpose_test.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)

configs/body/2d_kpt_sview_rgb_img/associative_embedding/crowdpose/higherhrnet_w32_crowdpose_640x640_udp.py

+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/crowdpose.py'
+]
+checkpoint_config = dict(interval=50)
+evaluation = dict(interval=50, metric='mAP', save_best='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 = 300
+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=640,
+    base_size=320,
+    base_sigma=2,
+    heatmap_size=[160, 320],
+    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='AssociativeEmbedding',
+    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='AEHigherResolutionHead',
+        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],
+        loss_keypoint=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_cfg=dict(),
+    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=False,
+        align_corners=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,
+        use_udp=True))
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='BottomUpRandomAffine',
+        rot_factor=30,
+        scale_factor=[0.75, 1.5],
+        scale_type='short',
+        trans_factor=40,
+        use_udp=True),
+    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,
+        use_udp=True,
+    ),
+    dict(
+        type='Collect',
+        keys=['img', 'joints', 'targets', 'masks'],
+        meta_keys=[]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='BottomUpGetImgSize', test_scale_factor=[1], use_udp=True),
+    dict(
+        type='BottomUpResizeAlign',
+        transforms=[
+            dict(type='ToTensor'),
+            dict(
+                type='NormalizeTensor',
+                mean=[0.485, 0.456, 0.406],
+                std=[0.229, 0.224, 0.225]),
+        ],
+        use_udp=True),
+    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/crowdpose'
+data = dict(
+    workers_per_gpu=2,
+    train_dataloader=dict(samples_per_gpu=16),
+    val_dataloader=dict(samples_per_gpu=1),
+    test_dataloader=dict(samples_per_gpu=1),
+    train=dict(
+        type='BottomUpCrowdPoseDataset',
+        ann_file=f'{data_root}/annotations/mmpose_crowdpose_trainval.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='BottomUpCrowdPoseDataset',
+        ann_file=f'{data_root}/annotations/mmpose_crowdpose_test.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='BottomUpCrowdPoseDataset',
+        ann_file=f'{data_root}/annotations/mmpose_crowdpose_test.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)

configs/body/2d_kpt_sview_rgb_img/associative_embedding/crowdpose/higherhrnet_w48_crowdpose_512x512.py

+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/crowdpose.py'
+]
+checkpoint_config = dict(interval=50)
+evaluation = dict(interval=50, metric='mAP', save_best='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 = 300
+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='AssociativeEmbedding',
+    pretrained='https://download.openmmlab.com/mmpose/'
+    'pretrain_models/hrnet_w48-8ef0771d.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=(48, 96)),
+            stage3=dict(
+                num_modules=4,
+                num_branches=3,
+                block='BASIC',
+                num_blocks=(4, 4, 4),
+                num_channels=(48, 96, 192)),
+            stage4=dict(
+                num_modules=3,
+                num_branches=4,
+                block='BASIC',
+                num_blocks=(4, 4, 4, 4),
+                num_channels=(48, 96, 192, 384))),
+    ),
+    keypoint_head=dict(
+        type='AEHigherResolutionHead',
+        in_channels=48,
+        num_joints=14,
+        tag_per_joint=True,
+        extra=dict(final_conv_kernel=1, ),
+        num_deconv_layers=1,
+        num_deconv_filters=[48],
+        num_deconv_kernels=[4],
+        num_basic_blocks=4,
+        cat_output=[True],
+        with_ae_loss=[True, False],
+        loss_keypoint=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_cfg=dict(),
+    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,
+        align_corners=False,
+        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))
+
+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/crowdpose'
+data = dict(
+    workers_per_gpu=2,
+    train_dataloader=dict(samples_per_gpu=16),
+    val_dataloader=dict(samples_per_gpu=1),
+    test_dataloader=dict(samples_per_gpu=1),
+    train=dict(
+        type='BottomUpCrowdPoseDataset',
+        ann_file=f'{data_root}/annotations/mmpose_crowdpose_trainval.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='BottomUpCrowdPoseDataset',
+        ann_file=f'{data_root}/annotations/mmpose_crowdpose_test.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='BottomUpCrowdPoseDataset',
+        ann_file=f'{data_root}/annotations/mmpose_crowdpose_test.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)

configs/body/2d_kpt_sview_rgb_img/associative_embedding/crowdpose/higherhrnet_w48_crowdpose_512x512_udp.py

+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/crowdpose.py'
+]
+checkpoint_config = dict(interval=50)
+evaluation = dict(interval=50, metric='mAP', save_best='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 = 300
+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='AssociativeEmbedding',
+    pretrained='https://download.openmmlab.com/mmpose/'
+    'pretrain_models/hrnet_w48-8ef0771d.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=(48, 96)),
+            stage3=dict(
+                num_modules=4,
+                num_branches=3,
+                block='BASIC',
+                num_blocks=(4, 4, 4),
+                num_channels=(48, 96, 192)),
+            stage4=dict(
+                num_modules=3,
+                num_branches=4,
+                block='BASIC',
+                num_blocks=(4, 4, 4, 4),
+                num_channels=(48, 96, 192, 384))),
+    ),
+    keypoint_head=dict(
+        type='AEHigherResolutionHead',
+        in_channels=48,
+        num_joints=14,
+        tag_per_joint=True,
+        extra=dict(final_conv_kernel=1, ),
+        num_deconv_layers=1,
+        num_deconv_filters=[48],
+        num_deconv_kernels=[4],
+        num_basic_blocks=4,
+        cat_output=[True],
+        with_ae_loss=[True, False],
+        loss_keypoint=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_cfg=dict(),
+    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=False,
+        align_corners=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,
+        use_udp=True))
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(
+        type='BottomUpRandomAffine',
+        rot_factor=30,
+        scale_factor=[0.75, 1.5],
+        scale_type='short',
+        trans_factor=40,
+        use_udp=True),
+    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,
+        use_udp=True,
+    ),
+    dict(
+        type='Collect',
+        keys=['img', 'joints', 'targets', 'masks'],
+        meta_keys=[]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='BottomUpGetImgSize', test_scale_factor=[1], use_udp=True),
+    dict(
+        type='BottomUpResizeAlign',
+        transforms=[
+            dict(type='ToTensor'),
+            dict(
+                type='NormalizeTensor',
+                mean=[0.485, 0.456, 0.406],
+                std=[0.229, 0.224, 0.225])
+        ],
+        use_udp=True),
+    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/crowdpose'
+data = dict(
+    workers_per_gpu=2,
+    train_dataloader=dict(samples_per_gpu=16),
+    val_dataloader=dict(samples_per_gpu=1),
+    test_dataloader=dict(samples_per_gpu=1),
+    train=dict(
+        type='BottomUpCrowdPoseDataset',
+        ann_file=f'{data_root}/annotations/mmpose_crowdpose_trainval.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='BottomUpCrowdPoseDataset',
+        ann_file=f'{data_root}/annotations/mmpose_crowdpose_test.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='BottomUpCrowdPoseDataset',
+        ann_file=f'{data_root}/annotations/mmpose_crowdpose_test.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)

configs/body/2d_kpt_sview_rgb_img/associative_embedding/crowdpose/mobilenetv2_crowdpose_512x512.py

+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/crowdpose.py'
+]
+checkpoint_config = dict(interval=50)
+evaluation = dict(interval=50, metric='mAP', save_best='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 = 300
+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],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    num_scales=1,
+    scale_aware_sigma=False,
+)
+
+# model settings
+model = dict(
+    type='AssociativeEmbedding',
+    pretrained='mmcls://mobilenet_v2',
+    backbone=dict(type='MobileNetV2', widen_factor=1., out_indices=(7, )),
+    keypoint_head=dict(
+        type='AESimpleHead',
+        in_channels=1280,
+        num_joints=14,
+        tag_per_joint=True,
+        with_ae_loss=[True],
+        loss_keypoint=dict(
+            type='MultiLossFactory',
+            num_joints=14,
+            num_stages=1,
+            ae_loss_type='exp',
+            with_ae_loss=[True],
+            push_loss_factor=[0.001],
+            pull_loss_factor=[0.001],
+            with_heatmaps_loss=[True],
+            heatmaps_loss_factor=[1.0])),
+    train_cfg=dict(),
+    test_cfg=dict(
+        num_joints=channel_cfg['dataset_joints'],
+        max_num_people=30,
+        scale_factor=[1],
+        with_heatmaps=[True],
+        with_ae=[True],
+        project2image=True,
+        align_corners=False,
+        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))
+
+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/crowdpose'
+data = dict(
+    workers_per_gpu=2,
+    train_dataloader=dict(samples_per_gpu=24),
+    val_dataloader=dict(samples_per_gpu=1),
+    test_dataloader=dict(samples_per_gpu=1),
+    train=dict(
+        type='BottomUpCrowdPoseDataset',
+        ann_file=f'{data_root}/annotations/mmpose_crowdpose_trainval.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='BottomUpCrowdPoseDataset',
+        ann_file=f'{data_root}/annotations/mmpose_crowdpose_test.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='BottomUpCrowdPoseDataset',
+        ann_file=f'{data_root}/annotations/mmpose_crowdpose_test.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)