-
-
Notifications
You must be signed in to change notification settings - Fork 108
/
Copy pathREADME.md
59 lines (42 loc) · 11.2 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
## Introduction
<a href="https://github.com/openseg-group/OCNet.pytorch">Official Repo</a>
<a href="https://github.com/SegmentationBLWX/sssegmentation/blob/main/ssseg/modules/models/segmentors/ocrnet/ocrnet.py">Code Snippet</a>
<details>
<summary align="left"><a href="https://arxiv.org/pdf/1909.11065.pdf">OCRNet (ECCV'2020)</a></summary>
```latex
@article{yuan2019object,
title={Object-contextual representations for semantic segmentation},
author={Yuan, Yuhui and Chen, Xilin and Wang, Jingdong},
journal={arXiv preprint arXiv:1909.11065},
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 | 76.75% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ocrnet/ocrnet_resnet50os8_voc.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_resnet50os8_voc.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_resnet50os8_voc.log) |
| R-101-D8 | ImageNet-1k-224x224 | 512x512 | LR/POLICY/BS/EPOCH: 0.01/poly/16/60 | trainaug/val | 78.82% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ocrnet/ocrnet_resnet101os8_voc.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_resnet101os8_voc.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_resnet101os8_voc.log) |
| HRNetV2p-W18-Small | ImageNet-1k-224x224 | 512x512 | LR/POLICY/BS/EPOCH: 0.01/poly/16/60 | trainaug/val | 72.80% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ocrnet/ocrnet_hrnetv2w18s_voc.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_hrnetv2w18s_voc.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_hrnetv2w18s_voc.log) |
| HRNetV2p-W18 | ImageNet-1k-224x224 | 512x512 | LR/POLICY/BS/EPOCH: 0.01/poly/16/60 | trainaug/val | 75.80% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ocrnet/ocrnet_hrnetv2w18_voc.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_hrnetv2w18_voc.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_hrnetv2w18_voc.log) |
| HRNetV2p-W48 | ImageNet-1k-224x224 | 512x512 | LR/POLICY/BS/EPOCH: 0.01/poly/16/60 | trainaug/val | 77.60% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ocrnet/ocrnet_hrnetv2w48_voc.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_hrnetv2w48_voc.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_hrnetv2w48_voc.log) |
#### ADE20k
| Backbone | Pretrain | Crop Size | Schedule | Train/Eval Set | mIoU | Download |
| :-: | :-: | :-: | :-: | :-: | :-: | :-: |
| R-50-D8 | ImageNet-1k-224x224 | 512x512 | LR/POLICY/BS/EPOCH: 0.01/poly/16/130 | train/val | 42.47% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ocrnet/ocrnet_resnet50os8_ade20k.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_resnet50os8_ade20k.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_resnet50os8_ade20k.log) |
| R-101-D8 | ImageNet-1k-224x224 | 512x512 | LR/POLICY/BS/EPOCH: 0.01/poly/16/130 | train/val | 43.99% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ocrnet/ocrnet_resnet101os8_ade20k.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_resnet101os8_ade20k.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_resnet101os8_ade20k.log) |
| HRNetV2p-W18-Small | ImageNet-1k-224x224 | 512x512 | LR/POLICY/BS/EPOCH: 0.01/poly/16/130 | train/val | 38.04% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ocrnet/ocrnet_hrnetv2w18s_ade20k.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_hrnetv2w18s_ade20k.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_hrnetv2w18s_ade20k.log) |
| HRNetV2p-W18 | ImageNet-1k-224x224 | 512x512 | LR/POLICY/BS/EPOCH: 0.01/poly/16/130 | train/val | 39.85% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ocrnet/ocrnet_hrnetv2w18_ade20k.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_hrnetv2w18_ade20k.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_hrnetv2w18_ade20k.log) |
| HRNetV2p-W48 | ImageNet-1k-224x224 | 512x512 | LR/POLICY/BS/EPOCH: 0.01/poly/16/130 | train/val | 44.03% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ocrnet/ocrnet_hrnetv2w48_ade20k.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_hrnetv2w48_ade20k.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_hrnetv2w48_ade20k.log) |
#### CityScapes
| Backbone | Pretrain | Crop Size | Schedule | Train/Eval Set | mIoU | Download |
| :-: | :-: | :-: | :-: | :-: | :-: | :-: |
| R-50-D8 | ImageNet-1k-224x224 | 512x1024 | LR/POLICY/BS/EPOCH: 0.01/poly/8/440 | train/val | 79.40% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ocrnet/ocrnet_resnet50os8_cityscapes.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_resnet50os8_cityscapes.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_resnet50os8_cityscapes.log) |
| R-101-D8 | ImageNet-1k-224x224 | 512x1024 | LR/POLICY/BS/EPOCH: 0.01/poly/8/440 | train/val | 80.61% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ocrnet/ocrnet_resnet101os8_cityscapes.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_resnet101os8_cityscapes.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_resnet101os8_cityscapes.log) |
| HRNetV2p-W18-Small | ImageNet-1k-224x224 | 512x1024 | LR/POLICY/BS/EPOCH: 0.01/poly/8/440 | train/val | 79.30% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ocrnet/ocrnet_hrnetv2w18s_cityscapes.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_hrnetv2w18s_cityscapes.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_hrnetv2w18s_cityscapes.log) |
| HRNetV2p-W18 | ImageNet-1k-224x224 | 512x1024 | LR/POLICY/BS/EPOCH: 0.01/poly/8/440 | train/val | 80.58% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ocrnet/ocrnet_hrnetv2w18_cityscapes.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_hrnetv2w18_cityscapes.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_hrnetv2w18_cityscapes.log) |
| HRNetV2p-W48 | ImageNet-1k-224x224 | 512x1024 | LR/POLICY/BS/EPOCH: 0.01/poly/8/440 | train/val | 81.44% | [cfg](https://raw.githubusercontent.com/SegmentationBLWX/sssegmentation/main/ssseg/configs/ocrnet/ocrnet_hrnetv2w48_cityscapes.py) | [model](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_hrnetv2w48_cityscapes.pth) | [log](https://github.com/SegmentationBLWX/modelstore/releases/download/ssseg_ocrnet/ocrnet_hrnetv2w48_cityscapes.log) |
## More
You can also download the model weights from following sources:
- BaiduNetdisk: https://pan.baidu.com/s/1gD-NJJWOtaHCtB0qHE79rA with access code **s757**