6月22日，上海·源创会，云技术专场火热报名中

forked from Gitee 极速下载 / AlphaPose

加入 Gitee

与超过 1200万开发者一起发现、参与优秀开源项目，私有仓库也完全免费：）

克隆/下载

MODEL_ZOO.md 4.63 KB

# MODEL ZOO

## MSCOCO dataset

| Model                    | Backbone | Detector | Input Size |     AP     | Speed |  Download | Config | Training Log |  
|--------------------------|----------|----------|------------|------------|-------|-----------|--------|--------------|
|[Simple Baseline](../configs/coco/resnet/256x192_res50_lr1e-3_1x-simple.yaml)    | ResNet50           | YOLOv3 | 256x192            | 70.6        | 2.94 iter/s | [model](https://drive.google.com/open?id=1nxyfUbvWDaaT9eDM7Y31ScSVPlGy6gfw) | [cfg](../configs/coco/resnet/256x192_res50_lr1e-3_1x-simple.yaml)    | [log](logs/simple_res50_256x192.log) |
|[Fast Pose](../configs/coco/resnet/256x192_res50_lr1e-3_1x.yaml)    | ResNet50           | YOLOv3 | 256x192            | 72.0        | 3.54 iter/s | [model](https://drive.google.com/open?id=1kQhnMRURFiy7NsdS8EFL-8vtqEXOgECn) | [cfg](../configs/coco/resnet/256x192_res50_lr1e-3_1x.yaml)    | [log](logs/fast_res50_256x192.log) |
|[Fast Pose (DUC)](../configs/coco/resnet/256x192_res50_lr1e-3_1x-duc.yaml)    | ResNet50 - unshuffle           | YOLOv3 | 256x192            | 72.4        | 2.91 iter/s | [model](https://drive.google.com/open?id=1RlnL_YnnmgOM4L9L2szFdUnTjwptqtL-) | [cfg](../configs/coco/resnet/256x192_res50_lr1e-3_1x-duc.yaml)    | [log](logs/fast_421_res50-shuffle_256x192.log) |
|[HRNet](../configs/coco/hrnet/256x192_w32_lr1e-3.yaml)    | HRNet-W32           | YOLOv3 | 256x192            | 72.5        | 2.13 iter/s | [model](https://drive.google.com/open?id=1i63BPlOnp2vSjIZ7ni4Yp3RCPQwqe922) | [cfg](../configs/coco/hrnet/256x192_w32_lr1e-3.yaml)    | [log](logs/hrnet_w32_256x192.log) |
|[Fast Pose (DCN)](../configs/coco/resnet/256x192_res50_lr1e-3_2x-dcn.yaml)    | ResNet50 - dcn           | YOLOv3 | 256x192            | 72.8        | 2.94 iter/s | [model](https://drive.google.com/open?id=1zUz9YIk6eALCbZrukxD7kQ554nhi1pVv) | [cfg](../configs/coco/resnet/256x192_res50_lr1e-3_2x-dcn.yaml)    | [log](logs/fast_dcn_res50_256x192.log) |
|[Fast Pose (DUC)](../configs/coco/resnet/256x192_res152_lr1e-3_1x-duc.yaml)    | ResNet152           | YOLOv3 | 256x192            | 73.3        | 1.62 iter/s | [model](https://drive.google.com/open?id=1kfyedqyn8exjbbNmYq8XGd2EooQjPtF9) | [cfg](../configs/coco/resnet/256x192_res152_lr1e-3_1x-duc.yaml)    | [log](logs/fast_421_res152_256x192.log) |

#### Notes
- All models are trained on keypoint train 2017 images which contains at least one human with keypoint annotations (64115 images).
- The evaluation is done on COCO keypoint val 2017 (5000 images).
- Flip test is used by default.
- One TITAN XP is used for speed test, with `batch_size=64` in each iteration.
- Offline human detection results are used in speed test.
- `FastPose` is our own network design. Paper coming soon!

## [Halpe dataset](https://github.com/Fang-Haoshu/Halpe-FullBody) (26 keypoints)

| Model                    | Backbone | Detector | Input Size |     AP     | Speed |  Download | Config | Training Log |  
|--------------------------|----------|----------|------------|------------|-------|-----------|--------|--------------|
|[Fast Pose](../configs/halpe_26/resnet/256x192_res50_lr1e-3_1x.yaml)    | ResNet50           | YOLOv3 | 256x192            | 69.0        | 3.54 iter/s | [Google](https://drive.google.com/file/d/1S-ROA28de-1zvLv-hVfPFJ5tFBYOSITb/view?usp=sharing) [Baidu](https://pan.baidu.com/s/1lvzMhoYgS6o6n8lVDx3GtQ) | [cfg](../configs/halpe_26/resnet/256x192_res50_lr1e-3_1x.yaml)    | [log]() |

You can run with:
```
python scripts/demo_inference.py --cfg configs/halpe_26/resnet/256x192_res50_lr1e-3_1x.yaml --checkpoint pretrained_models/halpe26_fast_res50_256x192.pth --indir examples/demo/ --save_img
```

#### Notes
- More models coming soon!

## [Halpe dataset](https://github.com/Fang-Haoshu/Halpe-FullBody) (136 keypoints)

| Model                    | Backbone | Detector | Input Size |     AP     | Speed |  Download | Config | Training Log |  
|--------------------------|----------|----------|------------|------------|-------|-----------|--------|--------------|
|[Fast Pose](../configs/halpe_136/resnet/256x192_res50_lr1e-3_1x.yaml)    | ResNet50           | YOLOv3 | 256x192            | 69.0        | 3.54 iter/s | [Google](https://drive.google.com/file/d/17vnGsMDbG4rf50kyj586BVJsiAspQv5v/view?usp=sharing) [Baidu](https://pan.baidu.com/s/1--9DsFjTyQrTMwsMjY7FGg) | [cfg](../configs/halpe_136/resnet/256x192_res50_lr1e-3_2x-regression.yaml)    | [log]() |

You can run with:
```
python scripts/demo_inference.py --cfg configs/halpe_136/resnet/256x192_res50_lr1e-3_2x-regression.yaml --checkpoint pretrained_models/halpe136_fast_res50_256x192.pth --indir examples/demo/ --save_img
```

#### Notes
- More models coming soon!

一键复制原始数据按行查看历史

提交于 2020-12-22 21:38 . Update MODEL_ZOO.md

MODEL ZOO

MSCOCO dataset

Model	Backbone	Detector	Input Size	AP	Speed	Download	Config	Training Log
Simple Baseline	ResNet50	YOLOv3	256x192	70.6	2.94 iter/s	model	cfg	log
Fast Pose	ResNet50	YOLOv3	256x192	72.0	3.54 iter/s	model	cfg	log
Fast Pose (DUC)	ResNet50 - unshuffle	YOLOv3	256x192	72.4	2.91 iter/s	model	cfg	log
HRNet	HRNet-W32	YOLOv3	256x192	72.5	2.13 iter/s	model	cfg	log
Fast Pose (DCN)	ResNet50 - dcn	YOLOv3	256x192	72.8	2.94 iter/s	model	cfg	log
Fast Pose (DUC)	ResNet152	YOLOv3	256x192	73.3	1.62 iter/s	model	cfg	log

Notes

All models are trained on keypoint train 2017 images which contains at least one human with keypoint annotations (64115 images).
The evaluation is done on COCO keypoint val 2017 (5000 images).
Flip test is used by default.
One TITAN XP is used for speed test, with batch_size=64 in each iteration.
Offline human detection results are used in speed test.
FastPose is our own network design. Paper coming soon!

Halpe dataset (26 keypoints)

Model	Backbone	Detector	Input Size	AP	Speed	Download	Config	Training Log
Fast Pose	ResNet50	YOLOv3	256x192	69.0	3.54 iter/s	Google Baidu	cfg	log

You can run with:

python scripts/demo_inference.py --cfg configs/halpe_26/resnet/256x192_res50_lr1e-3_1x.yaml --checkpoint pretrained_models/halpe26_fast_res50_256x192.pth --indir examples/demo/ --save_img

Notes

More models coming soon!

Halpe dataset (136 keypoints)

Model	Backbone	Detector	Input Size	AP	Speed	Download	Config	Training Log
Fast Pose	ResNet50	YOLOv3	256x192	69.0	3.54 iter/s	Google Baidu	cfg	log

You can run with:

python scripts/demo_inference.py --cfg configs/halpe_136/resnet/256x192_res50_lr1e-3_2x-regression.yaml --checkpoint pretrained_models/halpe136_fast_res50_256x192.pth --indir examples/demo/ --save_img

Notes

More models coming soon!

马建仓 AI 助手

尝试更多

代码解读

代码找茬

代码优化

Python

1

https://gitee.com/yangruoyang9987/AlphaPose.git

git@gitee.com:yangruoyang9987/AlphaPose.git

yangruoyang9987

AlphaPose

AlphaPose

master