-
-
Notifications
You must be signed in to change notification settings - Fork 108
/
Copy pathREADME.md
95 lines (65 loc) · 21.1 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
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
## Introduction
<a href="https://github.com/SegmentationBLWX/sssegmentation">Official Repo</a>
<a href="https://github.com/SegmentationBLWX/sssegmentation/blob/main/ssseg/modules/models/segmentors/idrnet/idrnet.py">Code Snippet</a>
<details>
<summary align="left"><a href="https://arxiv.org/pdf/2310.10755.pdf">IDRNet (NeurIPS'2023)</a></summary>
```latex
@inproceedings{jin2023idrnet,
title={IDRNet: Intervention-Driven Relation Network for Semantic Segmentation},
author={Jin, Zhenchao and Hu, Xiaowei and Zhu, Lingting and Song, Luchuan and Yuan, Li and Yu, Lequan},
booktitle={Thirty-Seventh Conference on Neural Information Processing Systems},
year={2023}
}
```
</details>
## Results of Different Frameworks
#### LIP
| Segmentor | Pretrain | Backbone | Crop Size | Schedule | Train/Eval Set | mIoU | Download |
| :-: | :-: | :-: | :-: | :-: | :-: | :-: | :-: |
| FCN | ImageNet-1k-224x224 | R-50-D8 | 473x473 | LR/POLICY/BS/EPOCH: 0.01/poly/32/150 | train/val | 51.24% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/idrnet/idrnet_fcn_resnet50os8_lip.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_fcn_resnet50os8_lip.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_fcn_resnet50os8_lip.log) |
| PSNet | ImageNet-1k-224x224 | R-50-D8 | 473x473 | LR/POLICY/BS/EPOCH: 0.01/poly/32/150 | train/val | 53.29% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/idrnet/idrnet_ppm_resnet50os8_lip.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_ppm_resnet50os8_lip.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_ppm_resnet50os8_lip.log) |
| UperNet | ImageNet-1k-224x224 | R-50-D8 | 473x473 | LR/POLICY/BS/EPOCH: 0.01/poly/32/150 | train/val | 54.00% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/idrnet/idrnet_upernet_resnet50os8_lip.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_upernet_resnet50os8_lip.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_upernet_resnet50os8_lip.log) |
| DeepLabV3 | ImageNet-1k-224x224 | R-50-D8 | 473x473 | LR/POLICY/BS/EPOCH: 0.01/poly/32/150 | train/val | 53.87% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/idrnet/idrnet_aspp_resnet50os8_lip.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_aspp_resnet50os8_lip.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_aspp_resnet50os8_lip.log) |
#### ADE20k
| Segmentor | Pretrain | Backbone | Crop Size | Schedule | Train/Eval Set | mIoU | Download |
| :-: | :-: | :-: | :-: | :-: | :-: | :-: | :-: |
| FCN | ImageNet-1k-224x224 | R-50-D8 | 512x512 | LR/POLICY/BS/EPOCH: 0.01/poly/16/130 | train/val | 43.61% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/idrnet/idrnet_fcn_resnet50os8_ade20k.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_fcn_resnet50os8_ade20k.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_fcn_resnet50os8_ade20k.log) |
| PSNet | ImageNet-1k-224x224 | R-50-D8 | 512x512 | LR/POLICY/BS/EPOCH: 0.01/poly/16/130 | train/val | 44.02% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/idrnet/idrnet_ppm_resnet50os8_ade20k.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_ppm_resnet50os8_ade20k.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_ppm_resnet50os8_ade20k.log) |
| UperNet | ImageNet-1k-224x224 | R-50-D8 | 512x512 | LR/POLICY/BS/EPOCH: 0.01/poly/16/130 | train/val | 44.84% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/idrnet/idrnet_upernet_resnet50os8_ade20k.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_upernet_resnet50os8_ade20k.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_upernet_resnet50os8_ade20k.log) |
| DeepLabV3 | ImageNet-1k-224x224 | R-50-D8 | 512x512 | LR/POLICY/BS/EPOCH: 0.01/poly/16/130 | train/val | 44.75% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/idrnet/idrnet_aspp_resnet50os8_ade20k.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_aspp_resnet50os8_ade20k.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_aspp_resnet50os8_ade20k.log) |
#### CityScapes
| Segmentor | Pretrain | Backbone | Crop Size | Schedule | Train/Eval Set | mIoU | Download |
| :-: | :-: | :-: | :-: | :-: | :-: | :-: | :-: |
| FCN | ImageNet-1k-224x224 | R-50-D8 | 512x1024 | LR/POLICY/BS/EPOCH: 0.01/poly/8/220 | train/val | 79.91% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/idrnet/idrnet_fcn_resnet50os8_cityscapes.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_fcn_resnet50os8_cityscapes.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_fcn_resnet50os8_cityscapes.log) |
| PSNet | ImageNet-1k-224x224 | R-50-D8 | 512x1024 | LR/POLICY/BS/EPOCH: 0.01/poly/8/220 | train/val | 79.88% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/idrnet/idrnet_ppm_resnet50os8_cityscapes.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_ppm_resnet50os8_cityscapes.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_ppm_resnet50os8_cityscapes.log) |
| UperNet | ImageNet-1k-224x224 | R-50-D8 | 512x1024 | LR/POLICY/BS/EPOCH: 0.01/poly/8/220 | train/val | 80.81% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/idrnet/idrnet_upernet_resnet50os8_cityscapes.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_upernet_resnet50os8_cityscapes.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_upernet_resnet50os8_cityscapes.log) |
| DeepLabV3 | ImageNet-1k-224x224 | R-50-D8 | 512x1024 | LR/POLICY/BS/EPOCH: 0.01/poly/8/220 | train/val | 80.69% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/idrnet/idrnet_aspp_resnet50os8_cityscapes.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_aspp_resnet50os8_cityscapes.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_aspp_resnet50os8_cityscapes.log) |
#### COCOStuff-10k
| Segmentor | Pretrain | Backbone | Crop Size | Schedule | Train/Eval Set | mIoU | Download |
| :-: | :-: | :-: | :-: | :-: | :-: | :-: | :-: |
| FCN | ImageNet-1k-224x224 | R-50-D8 | 512x512 | LR/POLICY/BS/EPOCH: 0.001/poly/16/110 | train/test | 38.61% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/idrnet/idrnet_fcn_resnet50os8_cocostuff10k.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_fcn_resnet50os8_cocostuff10k.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_fcn_resnet50os8_cocostuff10k.log) |
| PSNet | ImageNet-1k-224x224 | R-50-D8 | 512x512 | LR/POLICY/BS/EPOCH: 0.001/poly/16/110 | train/test | 39.13% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/idrnet/idrnet_ppm_resnet50os8_cocostuff10k.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_ppm_resnet50os8_cocostuff10k.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_ppm_resnet50os8_cocostuff10k.log) |
| UperNet | ImageNet-1k-224x224 | R-50-D8 | 512x512 | LR/POLICY/BS/EPOCH: 0.001/poly/16/110 | train/test | 39.35% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/idrnet/idrnet_upernet_resnet50os8_cocostuff10k.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_upernet_resnet50os8_cocostuff10k.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_upernet_resnet50os8_cocostuff10k.log) |
| DeepLabV3 | ImageNet-1k-224x224 | R-50-D8 | 512x512 | LR/POLICY/BS/EPOCH: 0.001/poly/16/110 | train/test | 39.31% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/idrnet/idrnet_aspp_resnet50os8_cocostuff10k.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_aspp_resnet50os8_cocostuff10k.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_aspp_resnet50os8_cocostuff10k.log) |
## SOTA Results
#### PASCAL-Context-59
| Segmentor | Pretrain | Backbone | Crop Size | Schedule | Train/Eval Set | mIoU/mIoU (ms+flip) | Download |
| :-: | :-: | :-: | :-: | :-: | :-: | :-: | :-: |
| UperNet | ImageNet-22k-384x384 | Swin-Large | 640x640 | LR/POLICY/BS/EPOCH: 0.00006/poly/16/260 | train/val | 63.82%/64.50% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/idrnet/idrnet_upernet_swinlarge_pascalcontext59.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_upernet_swinlarge_pascalcontext59.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_upernet_swinlarge_pascalcontext59.log) |
#### LIP
| Segmentor | Pretrain | Backbone | Crop Size | Schedule | Train/Eval Set | mIoU/mIoU (flip)/mIoU (ms+flip) | Download |
| :-: | :-: | :-: | :-: | :-: | :-: | :-: | :-: |
| UperNet | ImageNet-22k-384x384 | Swin-Large | 473x473 | LR/POLICY/BS/EPOCH: 0.00006/poly/16/110 | train/val | 60.53%/60.83%/61.17% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/idrnet/idrnet_upernet_swinlarge_lip.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_upernet_swinlarge_lip.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_upernet_swinlarge_lip.log) |
#### ADE20k
| Segmentor | Pretrain | Backbone | Crop Size | Schedule | Train/Eval Set | mIoU/mIoU (ms+flip) | Download |
| :-: | :-: | :-: | :-: | :-: | :-: | :-: | :-: |
| UperNet | ImageNet-22k-384x384 | Swin-Large | 640x640 | LR/POLICY/BS/EPOCH: 0.00006/poly/16/130 | train/val | 53.97%/54.68% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/idrnet/idrnet_upernet_swinlarge_ade20k.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_upernet_swinlarge_ade20k.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_upernet_swinlarge_ade20k.log) |
#### COCOStuff-10k
| Segmentor | Pretrain | Backbone | Crop Size | Schedule | Train/Eval Set | mIoU/mIoU (ms+flip) | Download |
| :-: | :-: | :-: | :-: | :-: | :-: | :-: | :-: |
| UperNet | ImageNet-22k-384x384 | Swin-Large | 640x640 | LR/POLICY/BS/EPOCH: 0.00006/poly/16/110 | train/test | 49.94%/50.54% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/idrnet/idrnet_upernet_swinlarge_cocostuff10k.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_upernet_swinlarge_cocostuff10k.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_idrnet/idrnet_upernet_swinlarge_cocostuff10k.log) |
## More
You can also download the model weights from following sources:
- BaiduNetdisk: https://pan.baidu.com/s/1gD-NJJWOtaHCtB0qHE79rA with access code **s757**
Please note that, due to differences in computational precision, the numerical values obtained when testing model performance on different versions of PyTorch or graphics cards may vary slightly.
This is a normal phenomenon and the performance differences are generally within 0.1%.