-
-
Notifications
You must be signed in to change notification settings - Fork 108
/
README.md
67 lines (48 loc) · 12.2 KB
/
README.md
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
## Introduction
<a href="https://github.com/liutinglt/CE2P">Official Repo</a>
<a href="https://github.com/SegmentationBLWX/sssegmentation/blob/main/ssseg/modules/models/segmentors/ce2p/ce2p.py">Code Snippet</a>
<details>
<summary align="left"><a href="https://arxiv.org/pdf/1809.05996.pdf">CE2P (AAAI'2019)</a></summary>
```latex
@inproceedings{ruan2019devil,
title={Devil in the details: Towards accurate single and multiple human parsing},
author={Ruan, Tao and Liu, Ting and Huang, Zilong and Wei, Yunchao and Wei, Shikui and Zhao, Yao},
booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
volume={33},
pages={4814--4821},
year={2019}
}
```
</details>
## Results
#### PASCAL VOC
| Backbone | Pretrain | Crop Size | Schedule | Train/Eval Set | mIoU | Download |
| :-: | :-: | :-: | :-: | :-: | :-: | :-: |
| R-50-D8 | ImageNet-1k-224x224 | 512x512 | LR/POLICY/BS/EPOCH: 0.01/poly/16/60 | trainaug/val | 75.69% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ce2p/ce2p_resnet50os8_voc.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet50os8_voc.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet50os8_voc.log) |
| R-50-D16 | ImageNet-1k-224x224 | 512x512 | LR/POLICY/BS/EPOCH: 0.01/poly/16/60 | trainaug/val | 74.58% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ce2p/ce2p_resnet50os16_voc.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet50os16_voc.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet50os16_voc.log) |
| R-101-D8 | ImageNet-1k-224x224 | 512x512 | LR/POLICY/BS/EPOCH: 0.01/poly/16/60 | trainaug/val | 77.77% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ce2p/ce2p_resnet101os8_voc.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet101os8_voc.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet101os8_voc.log) |
| R-101-D16 | ImageNet-1k-224x224 | 512x512 | LR/POLICY/BS/EPOCH: 0.01/poly/16/60 | trainaug/val | 76.84% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ce2p/ce2p_resnet101os16_voc.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet101os16_voc.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet101os16_voc.log) |
#### LIP
| Backbone | Pretrain | Crop Size | Schedule | Train/Eval Set | mIoU | Download |
| :-: | :-: | :-: | :-: | :-: | :-: | :-: |
| R-50-D8 | ImageNet-1k-224x224 | 473x473 | LR/POLICY/BS/EPOCH: 0.01/poly/32/150 | train/val | 52.42% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ce2p/ce2p_resnet50os8_lip.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet50os8_lip.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet50os8_lip.log) |
| R-50-D16 | ImageNet-1k-224x224 | 473x473 | LR/POLICY/BS/EPOCH: 0.01/poly/32/150 | train/val | 51.98% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ce2p/ce2p_resnet50os16_lip.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet50os16_lip.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet50os16_lip.log) |
| R-101-D8 | ImageNet-1k-224x224 | 473x473 | LR/POLICY/BS/EPOCH: 0.01/poly/32/150 | train/val | 54.79% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ce2p/ce2p_resnet101os8_lip.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet101os8_lip.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet101os8_lip.log) |
| R-101-D16 | ImageNet-1k-224x224 | 473x473 | LR/POLICY/BS/EPOCH: 0.01/poly/32/150 | train/val | 54.02% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ce2p/ce2p_resnet101os16_lip.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet101os16_lip.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet101os16_lip.log) |
#### CIHP
| Backbone | Pretrain | Crop Size | Schedule | Train/Eval Set | mIoU | Download |
| :-: | :-: | :-: | :-: | :-: | :-: | :-: |
| R-50-D8 | ImageNet-1k-224x224 | 473x473 | LR/POLICY/BS/EPOCH: 0.01/poly/32/150 | train/val | 61.15% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ce2p/ce2p_resnet50os8_cihp.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet50os8_cihp.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet50os8_cihp.log) |
| R-50-D16 | ImageNet-1k-224x224 | 473x473 | LR/POLICY/BS/EPOCH: 0.01/poly/32/150 | train/val | 60.15% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ce2p/ce2p_resnet50os16_cihp.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet50os16_cihp.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet50os16_cihp.log) |
| R-101-D8 | ImageNet-1k-224x224 | 473x473 | LR/POLICY/BS/EPOCH: 0.01/poly/32/150 | train/val | 63.83% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ce2p/ce2p_resnet101os8_cihp.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet101os8_cihp.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet101os8_cihp.log) |
| R-101-D16 | ImageNet-1k-224x224 | 473x473 | LR/POLICY/BS/EPOCH: 0.01/poly/32/150 | train/val | 62.25% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ce2p/ce2p_resnet101os16_cihp.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet101os16_cihp.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet101os16_cihp.log) |
#### ATR
| Backbone | Pretrain | Crop Size | Schedule | Train/Eval Set | mIoU | Download |
| :-: | :-: | :-: | :-: | :-: | :-: | :-: |
| R-50-D8 | ImageNet-1k-224x224 | 473x473 | LR/POLICY/BS/EPOCH: 0.01/poly/32/150 | train/val | 78.02% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ce2p/ce2p_resnet50os8_atr.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet50os8_atr.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet50os8_atr.log) |
| R-50-D16 | ImageNet-1k-224x224 | 473x473 | LR/POLICY/BS/EPOCH: 0.01/poly/32/150 | train/val | 77.62% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ce2p/ce2p_resnet50os16_atr.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet50os16_atr.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet50os16_atr.log) |
| R-101-D8 | ImageNet-1k-224x224 | 473x473 | LR/POLICY/BS/EPOCH: 0.01/poly/32/150 | train/val | 78.57% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ce2p/ce2p_resnet101os8_atr.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet101os8_atr.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet101os8_atr.log) |
| R-101-D16 | ImageNet-1k-224x224 | 473x473 | LR/POLICY/BS/EPOCH: 0.01/poly/32/150 | train/val | 78.25% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ce2p/ce2p_resnet101os16_atr.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet101os16_atr.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ce2p/ce2p_resnet101os16_atr.log) |
## More
You can also download the model weights from following sources:
- BaiduNetdisk: https://pan.baidu.com/s/1gD-NJJWOtaHCtB0qHE79rA with access code **s757**