Skip to content

Commit

Permalink
change name to mpii_trb (open-mmlab#48)
Browse files Browse the repository at this point in the history
  • Loading branch information
jin-s13 authored Aug 3, 2020
1 parent 1730a3d commit c01003e
Show file tree
Hide file tree
Showing 7 changed files with 63 additions and 40 deletions.
4 changes: 2 additions & 2 deletions configs/top_down/resnet/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -24,8 +24,8 @@
| pose_resnet_152 | 256x192 | 0.735 | 0.905 | 0.812 | 0.790 | 0.943 | [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 | 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 TRB-MPI val set.
### Results on MPII-TRB val set.

| Arch | Input Size | Skeleton Acc | Contour Acc | Mean Acc | ckpt | log |
| :--- | :--------: | :------: | :------: |:------: |:------: |:------: |
| pose_resnet_50 | 256x256 | 0.884 | 0.855 | 0.865 | [ckpt](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/top_down/resnet/res50_trbmpi_256x256-f0305d2e_20200727.pth) | [log](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/top_down/resnet/res50_trbmpi_256x256_20200727.log.json) |
| pose_resnet_50 | 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) |
Original file line number Diff line number Diff line change
Expand Up @@ -102,14 +102,14 @@
samples_per_gpu=32,
workers_per_gpu=2,
train=dict(
type='TopDownTRBMPIDataset',
ann_file=f'{data_root}/annotations/trb_mpi_train.json',
type='TopDownMpiiTrbDataset',
ann_file=f'{data_root}/annotations/mpii_trb_train.json',
img_prefix=f'{data_root}/images/',
data_cfg=data_cfg,
pipeline=train_pipeline),
val=dict(
type='TopDownTRBMPIDataset',
ann_file=f'{data_root}/annotations/trb_mpi_val.json',
type='TopDownMpiiTrbDataset',
ann_file=f'{data_root}/annotations/mpii_trb_val.json',
img_prefix=f'{data_root}/images/',
data_cfg=data_cfg,
pipeline=valid_pipeline),
Expand Down
73 changes: 47 additions & 26 deletions docs/getting_started.md
Original file line number Diff line number Diff line change
Expand Up @@ -8,8 +8,38 @@ For installation instructions, please see [install.md](install.md).
It is recommended to symlink the dataset root to `$MMPOSE/data`.
If your folder structure is different, you may need to change the corresponding paths in config files.

**For COCO data**, please download from [COCO download](http://cocodataset.org/#download), 2017 Train/Val is needed for COCO keypoints training and validation. [HRNet-Human-Pose-Estimation](https://github.com/HRNet/HRNet-Human-Pose-Estimation) provides person detection result of COCO val2017 to reproduce our multi-person pose estimation results. Please download from [OneDrive](https://1drv.ms/f/s!AhIXJn_J-blWzzDXoz5BeFl8sWM-)
Download and extract them under $MMPOSE/data, and make them look like this:

```
mmpose
├── mmpose
├── docs
├── tests
├── tools
├── configs
`── data
│── coco
│-- annotations
│ │-- person_keypoints_train2017.json
│ |-- person_keypoints_val2017.json
|-- person_detection_results
| |-- COCO_val2017_detections_AP_H_56_person.json
│-- train2017
│ │-- 000000000009.jpg
│ │-- 000000000025.jpg
│ │-- 000000000030.jpg
│ │-- ...
`-- val2017
│-- 000000000139.jpg
│-- 000000000285.jpg
│-- 000000000632.jpg
│-- ...
```

**For MPII data**, please download from [MPII Human Pose Dataset](http://human-pose.mpi-inf.mpg.de/).
We support both MPII dataset and TRB-MPI dataset. We have converted the original annotation files into json format, please download them from [mpii_annotations](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/datasets/mpii_annotations.tar.gz).
We have converted the original annotation files into json format, please download them from [mpii_annotations](https://openmmlab.oss-cn-hangzhou.aliyuncs.com/mmpose/datasets/mpii_annotations.tar).
Extract them under {MMPose}/data, and make them look like this:

```
Expand All @@ -22,21 +52,21 @@ mmpose
`── data
│── mpii
|── annotations
| |── trb_mpi_train.json
| |── trb_mpi_val.json
| |── gt_valid.mat
| |── test.json
| |── train.json
| |── trainval.json
| `── valid.json
| |── mpii_gt_valid.mat
| |── mpii_test.json
| |── mpii_train.json
| |── mpii_trainval.json
| `── mpii_valid.json
`── images
|── 000001163.jpg
|── 000003072.jpg
```

**For COCO data**, please download from [COCO download](http://cocodataset.org/#download), 2017 Train/Val is needed for COCO keypoints training and validation. [HRNet-Human-Pose-Estimation](https://github.com/HRNet/HRNet-Human-Pose-Estimation) provides person detection result of COCO val2017 to reproduce our multi-person pose estimation results. Please download from [OneDrive](https://1drv.ms/f/s!AhIXJn_J-blWzzDXoz5BeFl8sWM-)
Download and extract them under $MMPOSE/data, and make them look like this:

**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:

```
mmpose
Expand All @@ -46,22 +76,13 @@ mmpose
├── tools
├── configs
`── data
│── coco
│-- annotations
│ │-- person_keypoints_train2017.json
│ |-- person_keypoints_val2017.json
|-- person_detection_results
| |-- COCO_val2017_detections_AP_H_56_person.json
│-- train2017
│ │-- 000000000009.jpg
│ │-- 000000000025.jpg
│ │-- 000000000030.jpg
│ │-- ...
`-- val2017
│-- 000000000139.jpg
│-- 000000000285.jpg
│-- 000000000632.jpg
│-- ...
│── mpii
|── annotations
| |── trb_mpi_train.json
| |── trb_mpi_val.json
`── images
|── 000001163.jpg
|── 000003072.jpg
```

Expand Down
4 changes: 2 additions & 2 deletions mmpose/datasets/__init__.py
Original file line number Diff line number Diff line change
@@ -1,12 +1,12 @@
from .builder import build_dataloader, build_dataset
from .datasets import (BottomUpCocoDataset, TopDownCocoDataset,
TopDownTRBMPIDataset)
TopDownMpiiTrbDataset)
from .pipelines import Compose
from .registry import DATASETS, PIPELINES
from .samplers import DistributedSampler

__all__ = [
'TopDownCocoDataset', 'BottomUpCocoDataset', 'TopDownTRBMPIDataset',
'TopDownCocoDataset', 'BottomUpCocoDataset', 'TopDownMpiiTrbDataset',
'build_dataloader', 'build_dataset', 'Compose', 'DistributedSampler',
'DATASETS', 'PIPELINES'
]
6 changes: 4 additions & 2 deletions mmpose/datasets/datasets/__init__.py
Original file line number Diff line number Diff line change
@@ -1,4 +1,6 @@
from .bottom_up import BottomUpCocoDataset
from .top_down import TopDownCocoDataset, TopDownTRBMPIDataset
from .top_down import TopDownCocoDataset, TopDownMpiiTrbDataset

__all__ = ['TopDownCocoDataset', 'BottomUpCocoDataset', 'TopDownTRBMPIDataset']
__all__ = [
'TopDownCocoDataset', 'BottomUpCocoDataset', 'TopDownMpiiTrbDataset'
]
4 changes: 2 additions & 2 deletions mmpose/datasets/datasets/top_down/__init__.py
Original file line number Diff line number Diff line change
@@ -1,4 +1,4 @@
from .topdown_coco_dataset import TopDownCocoDataset
from .topdown_trbmpi_dataset import TopDownTRBMPIDataset
from .topdown_mpii_trb_dataset import TopDownMpiiTrbDataset

__all__ = ['TopDownCocoDataset', 'TopDownTRBMPIDataset']
__all__ = ['TopDownCocoDataset', 'TopDownMpiiTrbDataset']
Original file line number Diff line number Diff line change
Expand Up @@ -11,8 +11,8 @@


@DATASETS.register_module()
class TopDownTRBMPIDataset(TopDownBaseDataset):
"""CocoDataset dataset for top-down pose estimation.
class TopDownMpiiTrbDataset(TopDownBaseDataset):
"""MPII-TRB Dataset dataset for top-down pose estimation.
The dataset loads raw features and apply specified transforms
to return a dict containing the image tensors and other information.
Expand Down

0 comments on commit c01003e

Please sign in to comment.