add OneHand10K dataset

configs/top_down/resnet/README.md

@@ -29,3 +29,10 @@
 | Arch  | Input Size | Skeleton Acc   | Contour Acc   | Mean Acc | ckpt    | log     |
 | :--- | :--------: | :------: | :------: |:------: |:------: |:------: |
 | [pose_resnet_50](/configs/top_down/resnet/mpii_trb/res50_mpii_trb_256x256.py)  | 256x256 | 0.884 | 0.855 | 0.865 | [ckpt](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/top_down/resnet/res50_mpii_trb_256x256-f0305d2e_20200727.pth) | [log](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/top_down/resnet/res50_mpii_trb_256x256_20200727.log.json) |
+
+
+### Results on OneHand10K val set.
+
+| Arch  | Input Size | PCK@0.2 | ckpt    | log     |
+| :--- | :--------: | :------: | :------: |:------: |
+| [pose_resnet_50](/configs/top_down/resnet/onehand10k/res50_onehand10k_256x256.py)  | 256x256 | 0.985 | [ckpt](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/top_down/resnet/res50_onehand10k_256x256-e67998f6_20200813.pth) | [log](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/top_down/resnet/res50_onehand10k_256x256_20200813.log.json) |

configs/top_down/resnet/onehand10k/res50_onehand10k_256x256.py

@@ -0,0 +1,131 @@
+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='PCK')
+
+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=20,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook')
+    ])
+
+channel_cfg = dict(
+    num_output_channels=21,
+    dataset_joints=21,
+    dataset_channel=[
+        [
+            0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18,
+            19, 20
+        ],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19,
+        20
+    ])
+
+# 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=[256, 256],
+    heatmap_size=[64, 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'])
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=20, scale_factor=0.3),
+    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', 'flip_pairs'
+        ]),
+]
+
+valid_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', 'flip_pairs']),
+]
+
+test_pipeline = valid_pipeline
+
+data_root = 'data/onehand10k'
+data = dict(
+    samples_per_gpu=32,
+    workers_per_gpu=2,
+    train=dict(
+        type='TopDownOneHand10KDataset',
+        ann_file=f'{data_root}/annotations/onehand10k_train.json',
+        img_prefix=f'{data_root}/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline),
+    val=dict(
+        type='TopDownOneHand10KDataset',
+        ann_file=f'{data_root}/annotations/onehand10k_test.json',
+        img_prefix=f'{data_root}/',
+        data_cfg=data_cfg,
+        pipeline=valid_pipeline),
+    test=dict(
+        type='TopDownOneHand10KDataset',
+        ann_file=f'{data_root}/annotations/onehand10k_test.json',
+        img_prefix=f'{data_root}/',
+        data_cfg=data_cfg,
+        pipeline=valid_pipeline),
+)

docs/getting_started.md

@@ -86,6 +86,34 @@ mmpose
 
 ```
 
+**For OneHand10K data**, please download from [OneHand10K Dataset](https://www.yangangwang.com/papers/WANG-MCC-2018-10.html).
+Please download the annotation files from [onehand10k_annotations](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/datasets/onehand10k_annotations.tar).
+Extract them under {MMPose}/data, and make them look like this:
+
+```
+mmpose
+├── mmpose
+├── docs
+├── tests
+├── tools
+├── configs
+`── data
+    │── onehand10k
+        |── annotations
+        |   |── onehand10k_train.json
+        |   |── onehand10k_test.json
+        `── Train
+        |   |── source
+        |       |── 0.jpg
+        |       |── 1.jpg
+        |        ...
+        `── Test
+            |── source
+                |── 0.jpg
+                |── 1.jpg
+
+```
+
 For using custom datasets, please refer to [Tutorial 2: Adding New Dataset](tutorials/new_dataset.md)
 
 ## Prepare Pretrained Models

mmpose/datasets/__init__.py

@@ -1,12 +1,12 @@
 from .builder import build_dataloader, build_dataset
 from .datasets import (BottomUpCocoDataset, TopDownCocoDataset,
-                       TopDownMpiiTrbDataset)
+                       TopDownMpiiTrbDataset, TopDownOneHand10KDataset)
 from .pipelines import Compose
 from .registry import DATASETS, PIPELINES
 from .samplers import DistributedSampler
 
 __all__ = [
     'TopDownCocoDataset', 'BottomUpCocoDataset', 'TopDownMpiiTrbDataset',
-    'build_dataloader', 'build_dataset', 'Compose', 'DistributedSampler',
-    'DATASETS', 'PIPELINES'
+    'TopDownOneHand10KDataset', 'build_dataloader', 'build_dataset', 'Compose',
+    'DistributedSampler', 'DATASETS', 'PIPELINES'
 ]

mmpose/datasets/datasets/__init__.py

@@ -1,6 +1,8 @@
 from .bottom_up import BottomUpCocoDataset
-from .top_down import TopDownCocoDataset, TopDownMpiiTrbDataset
+from .top_down import (TopDownCocoDataset, TopDownMpiiTrbDataset,
+                       TopDownOneHand10KDataset)
 
 __all__ = [
-    'TopDownCocoDataset', 'BottomUpCocoDataset', 'TopDownMpiiTrbDataset'
+    'TopDownCocoDataset', 'BottomUpCocoDataset', 'TopDownMpiiTrbDataset',
+    'TopDownOneHand10KDataset'
 ]

mmpose/datasets/datasets/top_down/__init__.py

@@ -1,4 +1,7 @@
 from .topdown_coco_dataset import TopDownCocoDataset
 from .topdown_mpii_trb_dataset import TopDownMpiiTrbDataset
+from .topdown_onehand10k_dataset import TopDownOneHand10KDataset
 
-__all__ = ['TopDownCocoDataset', 'TopDownMpiiTrbDataset']
+__all__ = [
+    'TopDownCocoDataset', 'TopDownMpiiTrbDataset', 'TopDownOneHand10KDataset'
+]