diff --git a/configs/top_down/resnet/README.md b/configs/top_down/resnet/README.md
index 003ba6c2ef..9e4b3f27a3 100644
--- a/configs/top_down/resnet/README.md
+++ b/configs/top_down/resnet/README.md
@@ -25,6 +25,22 @@
 | [pose_resnet_152](/configs/top_down/resnet/coco/res152_coco_384x288.py) | 384x288 | 0.750 | 0.908 | 0.821 | 0.800 | 0.942 | [ckpt](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/top_down/resnet/res152_coco_384x288-3860d4c9_20200709.pth) | [log](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/top_down/resnet/res152_coco_384x288_20200709.log.json) |
 
 
+
+### Results on OCHuman test dataset with ground-truth bounding boxes
+
+Following the common setting, the models are trained on COCO train dataset, and evaluate on OCHuman dataset.
+
+| Arch  | Input Size | AP | AP<sup>50</sup> | AP<sup>75</sup> | AR | AR<sup>50</sup> | ckpt | log |
+| :-------------- | :-----------: | :------: | :------: | :------: | :------: | :------: |:------: |:------: |
+| [pose_resnet_50](/configs/top_down/resnet/coco/res50_coco_256x192.py)  | 256x192 | 0.546 | 0.726 | 0.593 | 0.592 | 0.755 | [ckpt](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/top_down/resnet/res50_coco_256x192-ec54d7f3_20200709.pth) | [log](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/top_down/resnet/res50_coco_256x192_20200709.log.json) |
+| [pose_resnet_50](/configs/top_down/resnet/coco/res50_coco_384x288.py)  | 384x288 | 0.539 | 0.723 | 0.574 | 0.588 | 0.756 | [ckpt](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/top_down/resnet/res50_coco_384x288-e6f795e9_20200709.pth) | [log](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/top_down/resnet/res50_coco_384x288_20200709.log.json) |
+| [pose_resnet_101](/configs/top_down/resnet/coco/res101_coco_256x192.py) | 256x192 | 0.559 | 0.724 | 0.606 | 0.605 | 0.751 | [ckpt](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/top_down/resnet/res101_coco_256x192-6e6babf0_20200708.pth) | [log](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/top_down/resnet/res101_coco_256x192_20200708.log.json) |
+| [pose_resnet_101](/configs/top_down/resnet/coco/res101_coco_384x288.py) | 384x288 | 0.571 | 0.715 | 0.615 | 0.615 | 0.748 | [ckpt](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/top_down/resnet/res101_coco_384x288-8c71bdc9_20200709.pth) | [log](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/top_down/resnet/res101_coco_384x288_20200709.log.json) |
+| [pose_resnet_152](/configs/top_down/resnet/coco/res152_coco_256x192.py) | 256x192 | 0.570 | 0.725 | 0.617 | 0.616 | 0.754 | [ckpt](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/top_down/resnet/res152_coco_256x192-f6e307c2_20200709.pth) | [log](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/top_down/resnet/res152_coco_256x192_20200709.log.json) |
+| [pose_resnet_152](/configs/top_down/resnet/coco/res152_coco_384x288.py) | 384x288 | 0.582 | 0.723 | 0.627 | 0.627 | 0.752 | [ckpt](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/top_down/resnet/res152_coco_384x288-3860d4c9_20200709.pth) | [log](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/top_down/resnet/res152_coco_384x288_20200709.log.json) |
+
+
+
 ### Results on MPII val set.
 
 | Arch  | Input Size | Mean | Mean@0.1   | ckpt    | log     |
diff --git a/configs/top_down/resnet/ochuman/res101_ochuman_256x192.py b/configs/top_down/resnet/ochuman/res101_ochuman_256x192.py
new file mode 100755
index 0000000000..8862b809b8
--- /dev/null
+++ b/configs/top_down/resnet/ochuman/res101_ochuman_256x192.py
@@ -0,0 +1,146 @@
+log_level = 'INFO'
+load_from = None
+resume_from = None
+dist_params = dict(backend='nccl')
+workflow = [('train', 1)]
+checkpoint_config = dict(interval=10)
+evaluation = dict(interval=5, metric='mAP')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+log_config = dict(
+    interval=50,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook')
+    ])
+
+channel_cfg = dict(
+    num_output_channels=17,
+    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
+    ])
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained='models/pytorch/imagenet/resnet101-5d3b4d8f.pth',
+    backbone=dict(type='ResNet', depth=101),
+    keypoint_head=dict(
+        type='TopDownSimpleHead',
+        in_channels=2048,
+        out_channels=channel_cfg['num_output_channels'],
+    ),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process=True,
+        shift_heatmap=True,
+        unbiased_decoding=False,
+        modulate_kernel=11),
+    loss_pose=dict(type='JointsMSELoss', use_target_weight=True))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    bbox_thr=1.0,
+    use_gt_bbox=True,
+    image_thr=0.0,
+    bbox_file='data/person_detection_results/'
+    'COCO_val2017_detections_AP_H_56_person.json',
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    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', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/ochuman'
+data = dict(
+    samples_per_gpu=64,
+    workers_per_gpu=2,
+    train=dict(
+        type='TopDownCocoDataset',
+        ann_file='data/coco/annotations/person_keypoints_train2017.json',
+        img_prefix='data/coco//train2017/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline),
+    val=dict(
+        type='TopDownOCHumanDataset',
+        ann_file=f'{data_root}/annotations/'
+        'ochuman_coco_format_val_range_0.00_1.00.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline),
+    test=dict(
+        type='TopDownOCHumanDataset',
+        ann_file=f'{data_root}/annotations/'
+        'ochuman_coco_format_test_range_0.00_1.00.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline),
+)
diff --git a/configs/top_down/resnet/ochuman/res101_ochuman_384x288.py b/configs/top_down/resnet/ochuman/res101_ochuman_384x288.py
new file mode 100755
index 0000000000..6addb9d1ab
--- /dev/null
+++ b/configs/top_down/resnet/ochuman/res101_ochuman_384x288.py
@@ -0,0 +1,146 @@
+log_level = 'INFO'
+load_from = None
+resume_from = None
+dist_params = dict(backend='nccl')
+workflow = [('train', 1)]
+checkpoint_config = dict(interval=10)
+evaluation = dict(interval=5, metric='mAP')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+log_config = dict(
+    interval=50,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook')
+    ])
+
+channel_cfg = dict(
+    num_output_channels=17,
+    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
+    ])
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained='models/pytorch/imagenet/resnet101-5d3b4d8f.pth',
+    backbone=dict(type='ResNet', depth=101),
+    keypoint_head=dict(
+        type='TopDownSimpleHead',
+        in_channels=2048,
+        out_channels=channel_cfg['num_output_channels'],
+    ),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process=True,
+        shift_heatmap=True,
+        unbiased_decoding=False,
+        modulate_kernel=11),
+    loss_pose=dict(type='JointsMSELoss', use_target_weight=True))
+
+data_cfg = dict(
+    image_size=[288, 384],
+    heatmap_size=[72, 96],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    bbox_thr=1.0,
+    use_gt_bbox=True,
+    image_thr=0.0,
+    bbox_file='data/person_detection_results/'
+    'COCO_val2017_detections_AP_H_56_person.json',
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=3),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    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', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/ochuman'
+data = dict(
+    samples_per_gpu=32,
+    workers_per_gpu=2,
+    train=dict(
+        type='TopDownCocoDataset',
+        ann_file='data/coco/annotations/person_keypoints_train2017.json',
+        img_prefix='data/coco//train2017/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline),
+    val=dict(
+        type='TopDownOCHumanDataset',
+        ann_file=f'{data_root}/annotations/'
+        'ochuman_coco_format_val_range_0.00_1.00.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline),
+    test=dict(
+        type='TopDownOCHumanDataset',
+        ann_file=f'{data_root}/annotations/'
+        'ochuman_coco_format_test_range_0.00_1.00.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline),
+)
diff --git a/configs/top_down/resnet/ochuman/res152_ochuman_256x192.py b/configs/top_down/resnet/ochuman/res152_ochuman_256x192.py
new file mode 100755
index 0000000000..094651f47a
--- /dev/null
+++ b/configs/top_down/resnet/ochuman/res152_ochuman_256x192.py
@@ -0,0 +1,146 @@
+log_level = 'INFO'
+load_from = None
+resume_from = None
+dist_params = dict(backend='nccl')
+workflow = [('train', 1)]
+checkpoint_config = dict(interval=10)
+evaluation = dict(interval=5, metric='mAP')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+log_config = dict(
+    interval=50,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook')
+    ])
+
+channel_cfg = dict(
+    num_output_channels=17,
+    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
+    ])
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained='models/pytorch/imagenet/resnet152-b121ed2d.pth',
+    backbone=dict(type='ResNet', depth=152),
+    keypoint_head=dict(
+        type='TopDownSimpleHead',
+        in_channels=2048,
+        out_channels=channel_cfg['num_output_channels'],
+    ),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process=True,
+        shift_heatmap=True,
+        unbiased_decoding=False,
+        modulate_kernel=11),
+    loss_pose=dict(type='JointsMSELoss', use_target_weight=True))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    bbox_thr=1.0,
+    use_gt_bbox=True,
+    image_thr=0.0,
+    bbox_file='data/person_detection_results/'
+    'COCO_val2017_detections_AP_H_56_person.json',
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    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', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/ochuman'
+data = dict(
+    samples_per_gpu=32,
+    workers_per_gpu=2,
+    train=dict(
+        type='TopDownCocoDataset',
+        ann_file='data/coco/annotations/person_keypoints_train2017.json',
+        img_prefix='data/coco//train2017/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline),
+    val=dict(
+        type='TopDownOCHumanDataset',
+        ann_file=f'{data_root}/annotations/'
+        'ochuman_coco_format_val_range_0.00_1.00.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline),
+    test=dict(
+        type='TopDownOCHumanDataset',
+        ann_file=f'{data_root}/annotations/'
+        'ochuman_coco_format_test_range_0.00_1.00.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline),
+)
diff --git a/configs/top_down/resnet/ochuman/res152_ochuman_384x288.py b/configs/top_down/resnet/ochuman/res152_ochuman_384x288.py
new file mode 100755
index 0000000000..a63c33d182
--- /dev/null
+++ b/configs/top_down/resnet/ochuman/res152_ochuman_384x288.py
@@ -0,0 +1,146 @@
+log_level = 'INFO'
+load_from = None
+resume_from = None
+dist_params = dict(backend='nccl')
+workflow = [('train', 1)]
+checkpoint_config = dict(interval=10)
+evaluation = dict(interval=5, metric='mAP')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+log_config = dict(
+    interval=50,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook')
+    ])
+
+channel_cfg = dict(
+    num_output_channels=17,
+    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
+    ])
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained='models/pytorch/imagenet/resnet152-b121ed2d.pth',
+    backbone=dict(type='ResNet', depth=152),
+    keypoint_head=dict(
+        type='TopDownSimpleHead',
+        in_channels=2048,
+        out_channels=channel_cfg['num_output_channels'],
+    ),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process=True,
+        shift_heatmap=True,
+        unbiased_decoding=False,
+        modulate_kernel=11),
+    loss_pose=dict(type='JointsMSELoss', use_target_weight=True))
+
+data_cfg = dict(
+    image_size=[288, 384],
+    heatmap_size=[72, 96],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    bbox_thr=1.0,
+    use_gt_bbox=True,
+    image_thr=0.0,
+    bbox_file='data/person_detection_results/'
+    'COCO_val2017_detections_AP_H_56_person.json',
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=3),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    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', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/ochuman'
+data = dict(
+    samples_per_gpu=48,
+    workers_per_gpu=2,
+    train=dict(
+        type='TopDownCocoDataset',
+        ann_file='data/coco/annotations/person_keypoints_train2017.json',
+        img_prefix='data/coco//train2017/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline),
+    val=dict(
+        type='TopDownOCHumanDataset',
+        ann_file=f'{data_root}/annotations/'
+        'ochuman_coco_format_val_range_0.00_1.00.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline),
+    test=dict(
+        type='TopDownOCHumanDataset',
+        ann_file=f'{data_root}/annotations/'
+        'ochuman_coco_format_test_range_0.00_1.00.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline),
+)
diff --git a/configs/top_down/resnet/ochuman/res50_ochuman_256x192.py b/configs/top_down/resnet/ochuman/res50_ochuman_256x192.py
new file mode 100644
index 0000000000..5d14360ef1
--- /dev/null
+++ b/configs/top_down/resnet/ochuman/res50_ochuman_256x192.py
@@ -0,0 +1,146 @@
+log_level = 'INFO'
+load_from = None
+resume_from = None
+dist_params = dict(backend='nccl')
+workflow = [('train', 1)]
+checkpoint_config = dict(interval=10)
+evaluation = dict(interval=1, metric='mAP')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+log_config = dict(
+    interval=50,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook')
+    ])
+
+channel_cfg = dict(
+    num_output_channels=17,
+    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
+    ])
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained='models/pytorch/imagenet/resnet50-19c8e357.pth',
+    backbone=dict(type='ResNet', depth=50),
+    keypoint_head=dict(
+        type='TopDownSimpleHead',
+        in_channels=2048,
+        out_channels=channel_cfg['num_output_channels'],
+    ),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process=True,
+        shift_heatmap=True,
+        unbiased_decoding=False,
+        modulate_kernel=11),
+    loss_pose=dict(type='JointsMSELoss', use_target_weight=True))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    bbox_thr=1.0,
+    use_gt_bbox=True,
+    image_thr=0.0,
+    bbox_file='data/coco/person_detection_results/'
+    'COCO_val2017_detections_AP_H_56_person.json',
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    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', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/ochuman'
+data = dict(
+    samples_per_gpu=64,
+    workers_per_gpu=2,
+    train=dict(
+        type='TopDownCocoDataset',
+        ann_file='data/coco/annotations/person_keypoints_train2017.json',
+        img_prefix='data/coco//train2017/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline),
+    val=dict(
+        type='TopDownOCHumanDataset',
+        ann_file=f'{data_root}/annotations/'
+        'ochuman_coco_format_val_range_0.00_1.00.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline),
+    test=dict(
+        type='TopDownOCHumanDataset',
+        ann_file=f'{data_root}/annotations/'
+        'ochuman_coco_format_test_range_0.00_1.00.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline),
+)
diff --git a/configs/top_down/resnet/ochuman/res50_ochuman_384x288.py b/configs/top_down/resnet/ochuman/res50_ochuman_384x288.py
new file mode 100755
index 0000000000..ddd6cf69dc
--- /dev/null
+++ b/configs/top_down/resnet/ochuman/res50_ochuman_384x288.py
@@ -0,0 +1,146 @@
+log_level = 'INFO'
+load_from = None
+resume_from = None
+dist_params = dict(backend='nccl')
+workflow = [('train', 1)]
+checkpoint_config = dict(interval=10)
+evaluation = dict(interval=5, metric='mAP')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+log_config = dict(
+    interval=50,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook')
+    ])
+
+channel_cfg = dict(
+    num_output_channels=17,
+    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
+    ])
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained='models/pytorch/imagenet/resnet50-19c8e357.pth',
+    backbone=dict(type='ResNet', depth=50),
+    keypoint_head=dict(
+        type='TopDownSimpleHead',
+        in_channels=2048,
+        out_channels=channel_cfg['num_output_channels'],
+    ),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process=True,
+        shift_heatmap=True,
+        unbiased_decoding=False,
+        modulate_kernel=11),
+    loss_pose=dict(type='JointsMSELoss', use_target_weight=True))
+
+data_cfg = dict(
+    image_size=[288, 384],
+    heatmap_size=[72, 96],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    bbox_thr=1.0,
+    use_gt_bbox=True,
+    image_thr=0.0,
+    bbox_file='data/person_detection_results/'
+    'COCO_val2017_detections_AP_H_56_person.json',
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=3),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    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', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/ochuman'
+data = dict(
+    samples_per_gpu=64,
+    workers_per_gpu=2,
+    train=dict(
+        type='TopDownCocoDataset',
+        ann_file='data/coco/annotations/person_keypoints_train2017.json',
+        img_prefix='data/coco//train2017/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline),
+    val=dict(
+        type='TopDownOCHumanDataset',
+        ann_file=f'{data_root}/annotations/'
+        'ochuman_coco_format_val_range_0.00_1.00.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline),
+    test=dict(
+        type='TopDownOCHumanDataset',
+        ann_file=f'{data_root}/annotations/'
+        'ochuman_coco_format_test_range_0.00_1.00.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline),
+)
diff --git a/docs/getting_started.md b/docs/getting_started.md
index 871d4e8bd0..7485b0e6f5 100644
--- a/docs/getting_started.md
+++ b/docs/getting_started.md
@@ -64,6 +64,29 @@ mmpose
 ```
 
 
+**For OCHuman data**, please download the images and annotations from [OCHuman](https://github.com/liruilong940607/OCHumanApi),
+Move them under $MMPOSE/data, and make them look like this:
+
+```
+mmpose
+├── mmpose
+├── docs
+├── tests
+├── tools
+├── configs
+`── data
+    │── ochuman
+        │-- annotations
+        │   │-- ochuman_coco_format_val_range_0.00_1.00.json
+        │   |-- ochuman_coco_format_test_range_0.00_1.00.json
+        |-- images
+            │-- 000001.jpg
+            │-- 000002.jpg
+            │-- 000003.jpg
+            │-- ...
+
+```
+
 **For MPII-TRB data**, please download from [MPII Human Pose Dataset](http://human-pose.mpi-inf.mpg.de/).
 Please download the annotation files from [mpii_trb_annotations](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/datasets/mpii_trb_annotations.tar).
 Extract them under {MMPose}/data, and make them look like this:
diff --git a/mmpose/datasets/__init__.py b/mmpose/datasets/__init__.py
index bb4f926cb9..aba648c7d1 100644
--- a/mmpose/datasets/__init__.py
+++ b/mmpose/datasets/__init__.py
@@ -1,13 +1,14 @@
 from .builder import build_dataloader, build_dataset
 from .datasets import (BottomUpCocoDataset, TopDownCocoDataset,
                        TopDownMpiiDataset, TopDownMpiiTrbDataset,
-                       TopDownOneHand10KDataset)
+                       TopDownOCHumanDataset, TopDownOneHand10KDataset)
 from .pipelines import Compose
 from .registry import DATASETS, PIPELINES
 from .samplers import DistributedSampler
 
 __all__ = [
     'TopDownCocoDataset', 'BottomUpCocoDataset', 'TopDownMpiiTrbDataset',
-    'TopDownOneHand10KDataset', 'TopDownMpiiDataset', 'build_dataloader',
-    'build_dataset', 'Compose', 'DistributedSampler', 'DATASETS', 'PIPELINES'
+    'TopDownOneHand10KDataset', 'TopDownMpiiDataset', 'TopDownOCHumanDataset',
+    'build_dataloader', 'build_dataset', 'Compose', 'DistributedSampler',
+    'DATASETS', 'PIPELINES'
 ]
diff --git a/mmpose/datasets/datasets/__init__.py b/mmpose/datasets/datasets/__init__.py
index 645f8005a5..50dc58578d 100644
--- a/mmpose/datasets/datasets/__init__.py
+++ b/mmpose/datasets/datasets/__init__.py
@@ -1,8 +1,10 @@
 from .bottom_up import BottomUpCocoDataset
 from .top_down import (TopDownCocoDataset, TopDownMpiiDataset,
-                       TopDownMpiiTrbDataset, TopDownOneHand10KDataset)
+                       TopDownMpiiTrbDataset, TopDownOCHumanDataset,
+                       TopDownOneHand10KDataset)
 
 __all__ = [
     'TopDownCocoDataset', 'BottomUpCocoDataset', 'TopDownMpiiDataset',
-    'TopDownMpiiTrbDataset', 'TopDownOneHand10KDataset'
+    'TopDownMpiiTrbDataset', 'TopDownOneHand10KDataset',
+    'TopDownOCHumanDataset'
 ]
diff --git a/mmpose/datasets/datasets/top_down/__init__.py b/mmpose/datasets/datasets/top_down/__init__.py
index fd9fc77ee8..72d6c6b6d1 100644
--- a/mmpose/datasets/datasets/top_down/__init__.py
+++ b/mmpose/datasets/datasets/top_down/__init__.py
@@ -1,9 +1,10 @@
 from .topdown_coco_dataset import TopDownCocoDataset
 from .topdown_mpii_dataset import TopDownMpiiDataset
 from .topdown_mpii_trb_dataset import TopDownMpiiTrbDataset
+from .topdown_ochuman_dataset import TopDownOCHumanDataset
 from .topdown_onehand10k_dataset import TopDownOneHand10KDataset
 
 __all__ = [
     'TopDownCocoDataset', 'TopDownMpiiTrbDataset', 'TopDownMpiiDataset',
-    'TopDownOneHand10KDataset'
+    'TopDownOneHand10KDataset', 'TopDownOCHumanDataset'
 ]
diff --git a/mmpose/datasets/datasets/top_down/topdown_coco_dataset.py b/mmpose/datasets/datasets/top_down/topdown_coco_dataset.py
index 8d5873f9d3..eddbdedec5 100644
--- a/mmpose/datasets/datasets/top_down/topdown_coco_dataset.py
+++ b/mmpose/datasets/datasets/top_down/topdown_coco_dataset.py
@@ -16,6 +16,10 @@
 class TopDownCocoDataset(TopDownBaseDataset):
     """CocoDataset dataset for top-down pose estimation.
 
+    `Microsoft COCO: Common Objects in Context' ECCV'2014
+    More details can be found in the `paper
+    <https://arxiv.org/abs/1405.0312>`_ .
+
     The dataset loads raw features and apply specified transforms
     to return a dict containing the image tensors and other information.
 
diff --git a/mmpose/datasets/datasets/top_down/topdown_mpii_dataset.py b/mmpose/datasets/datasets/top_down/topdown_mpii_dataset.py
index ec7381e052..dc6ea1fa85 100644
--- a/mmpose/datasets/datasets/top_down/topdown_mpii_dataset.py
+++ b/mmpose/datasets/datasets/top_down/topdown_mpii_dataset.py
@@ -13,6 +13,10 @@
 class TopDownMpiiDataset(TopDownBaseDataset):
     """MPII Dataset for top-down pose estimation.
 
+    `2D Human Pose Estimation: New Benchmark and State of the Art Analysis'
+    CVPR'2014. More details can be found in the `paper
+    <http://human-pose.mpi-inf.mpg.de/contents/andriluka14cvpr.pdf>`_ .
+
     The dataset loads raw features and apply specified transforms
     to return a dict containing the image tensors and other information.
 
diff --git a/mmpose/datasets/datasets/top_down/topdown_mpii_trb_dataset.py b/mmpose/datasets/datasets/top_down/topdown_mpii_trb_dataset.py
index 53286ef19d..f0c4260873 100644
--- a/mmpose/datasets/datasets/top_down/topdown_mpii_trb_dataset.py
+++ b/mmpose/datasets/datasets/top_down/topdown_mpii_trb_dataset.py
@@ -14,6 +14,10 @@
 class TopDownMpiiTrbDataset(TopDownBaseDataset):
     """MPII-TRB Dataset dataset for top-down pose estimation.
 
+    `TRB: A Novel Triplet Representation for Understanding 2D Human Body'
+    ICCV'2019 More details can be found in the `paper
+    <https://arxiv.org/abs/1910.11535>`_ .
+
     The dataset loads raw features and apply specified transforms
     to return a dict containing the image tensors and other information.
 
diff --git a/mmpose/datasets/datasets/top_down/topdown_ochuman_dataset.py b/mmpose/datasets/datasets/top_down/topdown_ochuman_dataset.py
new file mode 100644
index 0000000000..5f3b55eeb3
--- /dev/null
+++ b/mmpose/datasets/datasets/top_down/topdown_ochuman_dataset.py
@@ -0,0 +1,270 @@
+import os
+from collections import OrderedDict, defaultdict
+
+import numpy as np
+from pycocotools.coco import COCO
+
+from ....core.post_processing import oks_nms, soft_oks_nms
+from ...registry import DATASETS
+from .topdown_coco_dataset import TopDownCocoDataset
+
+
+def _get_mapping_id_name(imgs):
+    """
+    Args:
+        imgs (dict): dict of image info.
+    Returns:
+        id2name (dict): mapping image id to name.
+        name2id (dict): mapping image name to id.
+    """
+    id2name = {}
+    name2id = {}
+    for image_id, image in imgs.items():
+        file_name = image['file_name']
+        id2name[image_id] = file_name
+        name2id[file_name] = image_id
+
+    return id2name, name2id
+
+
+@DATASETS.register_module()
+class TopDownOCHumanDataset(TopDownCocoDataset):
+    """OChuman dataset for top-down pose estimation.
+
+    `Pose2Seg: Detection Free Human Instance Segmentation' CVPR'2019
+    More details can be found in the `paper
+    <https://arxiv.org/abs/1803.10683>`_ .
+
+    "Occluded Human (OCHuman)" dataset contains 8110 heavily occluded
+    human instances within 4731 images. OCHuman dataset is designed for
+    validation and testing. To evaluate on OCHuman, the model should be
+    trained on COCO training set, and then test the robustness of the
+    model to occlusion using OCHuman.
+
+    OCHuman keypoint indexes (same as COCO)::
+
+        0: 'nose',
+        1: 'left_eye',
+        2: 'right_eye',
+        3: 'left_ear',
+        4: 'right_ear',
+        5: 'left_shoulder',
+        6: 'right_shoulder',
+        7: 'left_elbow',
+        8: 'right_elbow',
+        9: 'left_wrist',
+        10: 'right_wrist',
+        11: 'left_hip',
+        12: 'right_hip',
+        13: 'left_knee',
+        14: 'right_knee',
+        15: 'left_ankle',
+        16: 'right_ankle'
+
+    Args:
+        ann_file (str): Path to the annotation file.
+        img_prefix (str): Path to a directory where images are held.
+            Default: None.
+        data_cfg (dict): config
+        pipeline (list[dict | callable]): A sequence of data transforms.
+        test_mode (bool): Store True when building test or
+            validation dataset. Default: False.
+    """
+
+    def __init__(self,
+                 ann_file,
+                 img_prefix,
+                 data_cfg,
+                 pipeline,
+                 test_mode=False):
+        super(TopDownCocoDataset, self).__init__(
+            ann_file, img_prefix, data_cfg, pipeline, test_mode=test_mode)
+
+        self.use_gt_bbox = data_cfg['use_gt_bbox']
+        self.bbox_file = data_cfg['bbox_file']
+        self.image_thr = data_cfg['image_thr']
+
+        self.soft_nms = data_cfg['soft_nms']
+        self.nms_thr = data_cfg['nms_thr']
+        self.oks_thr = data_cfg['oks_thr']
+        self.vis_thr = data_cfg['vis_thr']
+        self.bbox_thr = data_cfg['bbox_thr']
+
+        self.ann_info['flip_pairs'] = [[1, 2], [3, 4], [5, 6], [7, 8], [9, 10],
+                                       [11, 12], [13, 14], [15, 16]]
+
+        self.ann_info['upper_body_ids'] = (0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10)
+        self.ann_info['lower_body_ids'] = (11, 12, 13, 14, 15, 16)
+
+        self.ann_info['use_different_joint_weights'] = False
+        self.ann_info['joint_weights'] = np.array(
+            [
+                1., 1., 1., 1., 1., 1., 1., 1.2, 1.2, 1.5, 1.5, 1., 1., 1.2,
+                1.2, 1.5, 1.5
+            ],
+            dtype=np.float32).reshape((self.ann_info['num_joints'], 1))
+
+        self.coco = COCO(ann_file)
+
+        cats = [
+            cat['name'] for cat in self.coco.loadCats(self.coco.getCatIds())
+        ]
+        self.classes = ['__background__'] + cats
+        self.num_classes = len(self.classes)
+        self._class_to_ind = dict(zip(self.classes, range(self.num_classes)))
+        self._class_to_coco_ind = dict(zip(cats, self.coco.getCatIds()))
+        self._coco_ind_to_class_ind = dict([(self._class_to_coco_ind[cls],
+                                             self._class_to_ind[cls])
+                                            for cls in self.classes[1:]])
+        self.image_set_index = self.coco.getImgIds()
+        self.num_images = len(self.image_set_index)
+        self.id2name, self.name2id = _get_mapping_id_name(self.coco.imgs)
+
+        self.db = self._get_db()
+
+        print(f'=> num_images: {self.num_images}')
+        print(f'=> load {len(self.db)} samples')
+
+    def _get_db(self):
+        """Load dataset."""
+        assert self.use_gt_bbox
+        gt_db = self._load_coco_keypoint_annotations()
+        return gt_db
+
+    def _load_coco_keypoint_annotation_kernel(self, index):
+        """load annotation from COCOAPI.
+
+        Note:
+            bbox:[x1, y1, w, h]
+        Args:
+            index: coco image id
+        Returns:
+            db entry
+        """
+        im_ann = self.coco.loadImgs(index)[0]
+        width = im_ann['width']
+        height = im_ann['height']
+        num_joints = self.ann_info['num_joints']
+
+        ann_ids = self.coco.getAnnIds(imgIds=index, iscrowd=False)
+        objs = self.coco.loadAnns(ann_ids)
+
+        # sanitize bboxes
+        valid_objs = []
+        for obj in objs:
+            x, y, w, h = obj['bbox']
+            x1 = max(0, x)
+            y1 = max(0, y)
+            x2 = min(width - 1, x1 + max(0, w - 1))
+            y2 = min(height - 1, y1 + max(0, h - 1))
+            if obj['area'] > 0 and x2 >= x1 and y2 >= y1:
+                obj['clean_bbox'] = [x1, y1, x2 - x1, y2 - y1]
+                valid_objs.append(obj)
+        objs = valid_objs
+
+        rec = []
+        for obj in objs:
+            if max(obj['keypoints']) == 0:
+                continue
+            joints_3d = np.zeros((num_joints, 3), dtype=np.float)
+            joints_3d_visible = np.zeros((num_joints, 3), dtype=np.float)
+
+            keypoints = np.array(obj['keypoints']).reshape(-1, 3)
+            joints_3d[:, :2] = keypoints[:, :2]
+            joints_3d_visible[:, :2] = np.minimum(1, keypoints[:, 2:3])
+
+            center, scale = self._xywh2cs(*obj['clean_bbox'][:4])
+
+            image_file = os.path.join(self.img_prefix, self.id2name[index])
+            rec.append({
+                'image_file': image_file,
+                'center': center,
+                'scale': scale,
+                'rotation': 0,
+                'joints_3d': joints_3d,
+                'joints_3d_visible': joints_3d_visible,
+                'dataset': 'ochuman',
+                'bbox_score': 1
+            })
+
+        return rec
+
+    def evaluate(self, outputs, res_folder, metric='mAP', **kwargs):
+        """Evaluate coco keypoint results. The pose prediction results will be
+        saved in `${res_folder}/result_keypoints.json`.
+
+        Note:
+            num_keypoints: K
+
+        Args:
+            outputs (list(preds, boxes, image_path))
+                :preds (np.ndarray[1,K,3]): The first two dimensions are
+                    coordinates, score is the third dimension of the array.
+                :boxes (np.ndarray[1,6]): [center[0], center[1], scale[0]
+                    , scale[1],area, score]
+                :image_path (list[str]): For example, [ '/', 'v','a', 'l',
+                    '2', '0', '1', '7', '/', '0', '0', '0', '0', '0',
+                    '0', '3', '9', '7', '1', '3', '3', '.', 'j', 'p', 'g']
+            res_folder (str): Path of directory to save the results.
+            metric (str): Metric to be performed. Defaults: 'mAP'.
+
+        Returns:
+            name_value (dict): Evaluation results for evaluation metric.
+        """
+        assert metric == 'mAP'
+
+        res_file = os.path.join(res_folder, 'result_keypoints.json')
+
+        kpts = defaultdict(list)
+        for preds, boxes, image_path in outputs:
+            str_image_path = ''.join(image_path)
+            image_id = self.name2id[os.path.basename(str_image_path)]
+
+            kpts[image_id].append({
+                'keypoints': preds[0],
+                'center': boxes[0][0:2],
+                'scale': boxes[0][2:4],
+                'area': boxes[0][4],
+                'score': boxes[0][5],
+                'image_id': image_id,
+            })
+
+        # rescoring and oks nms
+        num_joints = self.ann_info['num_joints']
+        vis_thr = self.vis_thr
+        oks_thr = self.oks_thr
+        oks_nmsed_kpts = []
+        for img in kpts.keys():
+            img_kpts = kpts[img]
+            for n_p in img_kpts:
+                box_score = n_p['score']
+                kpt_score = 0
+                valid_num = 0
+                for n_jt in range(0, num_joints):
+                    t_s = n_p['keypoints'][n_jt][2]
+                    if t_s > vis_thr:
+                        kpt_score = kpt_score + t_s
+                        valid_num = valid_num + 1
+                if valid_num != 0:
+                    kpt_score = kpt_score / valid_num
+                # rescoring
+                n_p['score'] = kpt_score * box_score
+
+            if self.soft_nms:
+                keep = soft_oks_nms(
+                    [img_kpts[i] for i in range(len(img_kpts))], oks_thr)
+            else:
+                keep = oks_nms([img_kpts[i] for i in range(len(img_kpts))],
+                               oks_thr)
+
+            if len(keep) == 0:
+                oks_nmsed_kpts.append(img_kpts)
+            else:
+                oks_nmsed_kpts.append([img_kpts[_keep] for _keep in keep])
+
+        self._write_coco_keypoint_results(oks_nmsed_kpts, res_file)
+
+        info_str = self._do_python_keypoint_eval(res_file)
+        name_value = OrderedDict(info_str)
+
+        return name_value
diff --git a/mmpose/datasets/datasets/top_down/topdown_onehand10k_dataset.py b/mmpose/datasets/datasets/top_down/topdown_onehand10k_dataset.py
index f7549283d2..7e83bc8113 100644
--- a/mmpose/datasets/datasets/top_down/topdown_onehand10k_dataset.py
+++ b/mmpose/datasets/datasets/top_down/topdown_onehand10k_dataset.py
@@ -14,6 +14,11 @@
 class TopDownOneHand10KDataset(TopDownBaseDataset):
     """OneHand10K dataset for top-down hand pose estimation.
 
+    `Mask-pose Cascaded CNN for 2D Hand Pose Estimation from
+    Single Color Images' TCSVT'2019
+    More details can be found in the `paper
+    <https://www.yangangwang.com/papers/WANG-MCC-2018-10.pdf>`_ .
+
     The dataset loads raw features and apply specified transforms
     to return a dict containing the image tensors and other information.