Lxinyang cd246e3635 [Feature] Support Twins (NeurIPS2021) (#989)
* debug

* debug

* debug

* this is a debug step, and needs to be recovered

* need recover

* git

* debug

* git

* git

* git

* git

* git

* git

* debug need recover

* debug

* git

* debug

* debug

* debug

* debug

* debug

* debug

* debug

* debug

* debugf

* debug

* debug

* debug

* debug

* debug

* debug

* debug

* debug

* git

* git

* git

* use config small/base/large

* debug

* debug

* git

* debug

* git

* debug

* debug

* debug args

* debug

* debug

* git

* git

* debug

* git

* git

* git

* git

* git

* debug

* debug

* git

* debug

* git

* debug

* debug

* debug

* debug

* git

* debug

* git

* git

* debug

* debug

* git

* git

* git

* git

* debug

* debug

* debug

* debug

* git

* debug

* debug

* git

* git

* debug

* debug

* git

* debug

* debug

* debug

* git

* debug

* debug

* debug

* Please enter the commit message for your changes. Lines starting

* git

* git

* debug

* debug

* debug

* git

* git

* debug

* debug

* debug

* debug

* debug

* debug

* debug

* debug

* debug

* debug

* debug

* git

* debug

* debug

* debug

* debug

* debug

* debug

* debug

* git

* fix pre-commit error

* fix error

* git

* git

* git

* git

* git

* git

* debug

* debug

* debug

* debug

* debug

* debug

* git

* debug

* debug

* debug

* debug

* debug

* debug

* debug

* debug

* debug

* git

* git

* git

* debug

* debug

* debug

* git

* git

* git

* git

* git

* git

* git

* git

* git

* debug

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* fix unittest error

* fix config errors

* fix twins2mmseg bug

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* git

* fix init_weights() in twins.py

* git

* git

* git

* git

* fix comment

* fix comment

* fix comment

* fix comment

* fix unit test coverage in TwinsPR

* Add Twins README

* Add Twins README

* twins refactor

* twins refactor

* delete init_cfg in FFN

* delete init_cfg in FFN

* Update mmseg/models/backbones/twins.py

* Update mmseg/models/backbones/twins.py

* Update mmseg/models/backbones/twins.py

Co-authored-by: Junjun2016 <hejunjun@sjtu.edu.cn>

* Update mmseg/models/backbones/twins.py

* add conference name

Co-authored-by: linxinyang <linxinyang@meituan.com>
Co-authored-by: MengzhangLI <mcmong@pku.edu.cn>
Co-authored-by: Junjun2016 <hejunjun@sjtu.edu.cn>
2021-12-09 19:18:10 +08:00

280 lines
9.0 KiB
YAML

Collections:
- Name: twins
Metadata:
Training Data:
- ADE20K
Paper:
URL: https://arxiv.org/pdf/2104.13840.pdf
Title: 'Twins: Revisiting the Design of Spatial Attention in Vision Transformers'
README: configs/twins/README.md
Code:
URL: https://github.com/open-mmlab/mmsegmentation/blob/v0.20.0/mmseg/models/backbones/twins.py#L352
Version: v0.20.0
Converted From:
Code: https://github.com/Meituan-AutoML/Twins
Models:
- Name: twins_pcpvt-s_fpn_fpnhead_8x4_512x512_80k_ade20k
In Collection: twins
Metadata:
backbone: PCPVT-S
crop size: (512,512)
lr schd: 80000
inference time (ms/im):
- value: 36.83
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 6.6
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 43.26
mIoU(ms+flip): 44.11
Config: configs/twins/twins_pcpvt-s_fpn_fpnhead_8x4_512x512_80k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-s_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_pcpvt-s_fpn_fpnhead_8x4_512x512_80k_ade20k_20211201_204132-41acd132.pth
- Name: twins_pcpvt-s_uperhead_8x4_512x512_160k_ade20k
In Collection: twins
Metadata:
backbone: PCPVT-S
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 70.22
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 9.67
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 46.04
mIoU(ms+flip): 46.92
Config: configs/twins/twins_pcpvt-s_uperhead_8x4_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-s_uperhead_8x4_512x512_160k_ade20k/twins_pcpvt-s_uperhead_8x4_512x512_160k_ade20k_20211201_233537-8e99c07a.pth
- Name: twins_pcpvt-b_fpn_fpnhead_8x4_512x512_80k_ade20k
In Collection: twins
Metadata:
backbone: PCPVT-B
crop size: (512,512)
lr schd: 80000
inference time (ms/im):
- value: 50.84
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 8.41
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 45.66
mIoU(ms+flip): 46.48
Config: configs/twins/twins_pcpvt-b_fpn_fpnhead_8x4_512x512_80k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-b_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_pcpvt-b_fpn_fpnhead_8x4_512x512_80k_ade20k_20211130_141019-d396db72.pth
- Name: twins_pcpvt-b_uperhead_8x2_512x512_160k_ade20k
In Collection: twins
Metadata:
backbone: PCPVT-B
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 83.06
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 6.46
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 47.91
mIoU(ms+flip): 48.64
Config: configs/twins/twins_pcpvt-b_uperhead_8x2_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-b_uperhead_8x2_512x512_160k_ade20k/twins_pcpvt-b_uperhead_8x2_512x512_160k_ade20k_20211130_141020-02094ea5.pth
- Name: twins_pcpvt-l_fpn_fpnhead_8x4_512x512_80k_ade20k
In Collection: twins
Metadata:
backbone: PCPVT-L
crop size: (512,512)
lr schd: 80000
inference time (ms/im):
- value: 69.83
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 10.78
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 45.94
mIoU(ms+flip): 46.7
Config: configs/twins/twins_pcpvt-l_fpn_fpnhead_8x4_512x512_80k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-l_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_pcpvt-l_fpn_fpnhead_8x4_512x512_80k_ade20k_20211201_105226-bc6d61dc.pth
- Name: twins_pcpvt-l_uperhead_8x2_512x512_160k_ade20k
In Collection: twins
Metadata:
backbone: PCPVT-L
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 93.46
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 7.82
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 49.35
mIoU(ms+flip): 50.08
Config: configs/twins/twins_pcpvt-l_uperhead_8x2_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-l_uperhead_8x2_512x512_160k_ade20k/twins_pcpvt-l_uperhead_8x2_512x512_160k_ade20k_20211201_075053-c6095c07.pth
- Name: twins_svt-s_fpn_fpnhead_8x4_512x512_80k_ade20k
In Collection: twins
Metadata:
backbone: SVT-S
crop size: (512,512)
lr schd: 80000
inference time (ms/im):
- value: 33.57
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 5.8
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 44.47
mIoU(ms+flip): 45.42
Config: configs/twins/twins_svt-s_fpn_fpnhead_8x4_512x512_80k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-s_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_svt-s_fpn_fpnhead_8x4_512x512_80k_ade20k_20211130_141006-0a0d3317.pth
- Name: twins_svt-s_uperhead_8x2_512x512_160k_ade20k
In Collection: twins
Metadata:
backbone: SVT-S
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 66.27
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 4.93
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 46.08
mIoU(ms+flip): 46.96
Config: configs/twins/twins_svt-s_uperhead_8x2_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-s_uperhead_8x2_512x512_160k_ade20k/twins_svt-s_uperhead_8x2_512x512_160k_ade20k_20211130_141005-e48a2d94.pth
- Name: twins_svt-b_fpn_fpnhead_8x4_512x512_80k_ade20k
In Collection: twins
Metadata:
backbone: SVT-B
crop size: (512,512)
lr schd: 80000
inference time (ms/im):
- value: 47.39
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 8.75
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 46.77
mIoU(ms+flip): 47.47
Config: configs/twins/twins_svt-b_fpn_fpnhead_8x4_512x512_80k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-b_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_svt-b_fpn_fpnhead_8x4_512x512_80k_ade20k_20211201_113849-88b2907c.pth
- Name: twins_svt-b_uperhead_8x2_512x512_160k_ade20k
In Collection: twins
Metadata:
backbone: SVT-B
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 78.99
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 6.77
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 48.04
mIoU(ms+flip): 48.87
Config: configs/twins/twins_svt-b_uperhead_8x2_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-b_uperhead_8x2_512x512_160k_ade20k/twins_svt-b_uperhead_8x2_512x512_160k_ade20k_20211202_040826-0943a1f1.pth
- Name: twins_svt-l_fpn_fpnhead_8x4_512x512_80k_ade20k
In Collection: twins
Metadata:
backbone: SVT-L
crop size: (512,512)
lr schd: 80000
inference time (ms/im):
- value: 56.18
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 11.2
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 46.55
mIoU(ms+flip): 47.74
Config: configs/twins/twins_svt-l_fpn_fpnhead_8x4_512x512_80k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-l_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_svt-l_fpn_fpnhead_8x4_512x512_80k_ade20k_20211130_141005-1d59bee2.pth
- Name: twins_svt-l_uperhead_8x2_512x512_160k_ade20k
In Collection: twins
Metadata:
backbone: SVT-L
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 93.2
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 8.41
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 49.65
mIoU(ms+flip): 50.63
Config: configs/twins/twins_svt-l_uperhead_8x2_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-l_uperhead_8x2_512x512_160k_ade20k/twins_svt-l_uperhead_8x2_512x512_160k_ade20k_20211130_141005-3e2cae61.pth