Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Feature] Support PointRend #109

Merged
merged 4 commits into from
Sep 7, 2020
+554 −4
Merged

[Feature] Support PointRend #109

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
3d841b5
[Feature] Support PointRend
xvjiarui Sep 6, 2020
e9cb01b
add previous test
xvjiarui Sep 6, 2020
590f72b
merge master
xvjiarui Sep 7, 2020
d66bfc3
update modelzoo
xvjiarui Sep 7, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -73,6 +73,7 @@ Supported methods:
- [x] [OCRNet](configs/ocrnet)
- [x] [Fast-SCNN](configs/fastscnn)
- [x] [Semantic FPN](configs/sem_fpn)
- [x] [PointRend](configs/point_rend)
- [x] [EMANet](configs/emanet)
- [x] [DNLNet](configs/dnlnet)
- [x] [Mixed Precision (FP16) Training](configs/fp16/README.md)
Expand Down
56 changes: 56 additions & 0 deletions configs/_base_/models/pointrend_r50.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,56 @@
# model settings
norm_cfg = dict(type='SyncBN', requires_grad=True)
model = dict(
type='CascadeEncoderDecoder',
num_stages=2,
pretrained='open-mmlab://resnet50_v1c',
backbone=dict(
type='ResNetV1c',
depth=50,
num_stages=4,
out_indices=(0, 1, 2, 3),
dilations=(1, 1, 1, 1),
strides=(1, 2, 2, 2),
norm_cfg=norm_cfg,
norm_eval=False,
style='pytorch',
contract_dilation=True),
neck=dict(
type='FPN',
in_channels=[256, 512, 1024, 2048],
out_channels=256,
num_outs=4),
decode_head=[
dict(
type='FPNHead',
in_channels=[256, 256, 256, 256],
in_index=[0, 1, 2, 3],
feature_strides=[4, 8, 16, 32],
channels=128,
dropout_ratio=-1,
num_classes=19,
norm_cfg=norm_cfg,
align_corners=False,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
dict(
type='PointHead',
in_channels=[256],
in_index=[0],
channels=256,
num_fcs=3,
coarse_pred_each_layer=True,
dropout_ratio=-1,
num_classes=19,
align_corners=False,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0))
])
# model training and testing settings
train_cfg = dict(
num_points=2048, oversample_ratio=3, importance_sample_ratio=0.75)
test_cfg = dict(
mode='whole',
subdivision_steps=2,
subdivision_num_points=8196,
scale_factor=2)
27 changes: 27 additions & 0 deletions configs/point_rend/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,27 @@
# PointRend: Image Segmentation as Rendering

## Introduction
```
@misc{alex2019pointrend,
title={PointRend: Image Segmentation as Rendering},
author={Alexander Kirillov and Yuxin Wu and Kaiming He and Ross Girshick},
year={2019},
eprint={1912.08193},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
```

## Results and models

### Cityscapes
| Method | Backbone | Crop Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | download |
|-----------|----------|-----------|--------:|---------:|----------------|------:|---------------|----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| PointRend | R-50 | 512x1024 | 80000 | 3.1 | 8.48 | 76.47 | 78.13 | [model](https://openmmlab.oss-accelerate.aliyuncs.com/mmsegmentation/v0.5/point_rend/pointrend_r50_512x1024_80k_cityscapes/pointrend_r50_512x1024_80k_cityscapes_20200711_015821-bb1ff523.pth) | [log](https://openmmlab.oss-accelerate.aliyuncs.com/mmsegmentation/v0.5/point_rend/pointrend_r50_512x1024_80k_cityscapes/pointrend_r50_512x1024_80k_cityscapes-20200715_214714.log.json) |
| PointRend | R-101 | 512x1024 | 80000 | 4.2 | 7.00 | 78.30 | 79.97 | [model](https://openmmlab.oss-accelerate.aliyuncs.com/mmsegmentation/v0.5/point_rend/pointrend_r101_512x1024_80k_cityscapes/pointrend_r101_512x1024_80k_cityscapes_20200711_170850-d0ca84be.pth) | [log](https://openmmlab.oss-accelerate.aliyuncs.com/mmsegmentation/v0.5/point_rend/pointrend_r101_512x1024_80k_cityscapes/pointrend_r101_512x1024_80k_cityscapes-20200715_214824.log.json) |

### ADE20K
| Method | Backbone | Crop Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | download |
|-----------|----------|-----------|--------:|---------:|----------------|------:|---------------|------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| PointRend | R-50 | 512x512 | 160000 | 5.1 | 17.31 | 37.64 | 39.17 | [model](https://openmmlab.oss-accelerate.aliyuncs.com/mmsegmentation/v0.5/point_rend/pointrend_r50_512x512_160k_ade20k/pointrend_r50_512x512_160k_ade20k_20200807_232644-ac3febf2.pth) | [log](https://openmmlab.oss-accelerate.aliyuncs.com/mmsegmentation/v0.5/point_rend/pointrend_r50_512x512_160k_ade20k/pointrend_r50_512x512_160k_ade20k-20200807_232644.log.json) |
| PointRend | R-101 | 512x512 | 160000 | 6.1 | 15.50 | 40.02 | 41.60 | [model](https://openmmlab.oss-accelerate.aliyuncs.com/mmsegmentation/v0.5/point_rend/pointrend_r101_512x512_160k_ade20k/pointrend_r101_512x512_160k_ade20k_20200808_030852-8834902a.pth) | [log](https://openmmlab.oss-accelerate.aliyuncs.com/mmsegmentation/v0.5/point_rend/pointrend_r101_512x512_160k_ade20k/pointrend_r101_512x512_160k_ade20k-20200808_030852.log.json) |
2 changes: 2 additions & 0 deletions configs/point_rend/pointrend_r101_512x1024_80k_cityscapes.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,2 @@
_base_ = './pointrend_r50_512x1024_80k_cityscapes.py'
model = dict(pretrained='open-mmlab://resnet101_v1c', backbone=dict(depth=101))
2 changes: 2 additions & 0 deletions configs/point_rend/pointrend_r101_512x512_160k_ade20k.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,2 @@
_base_ = './pointrend_r50_512x512_160k_ade20k.py'
model = dict(pretrained='open-mmlab://resnet101_v1c', backbone=dict(depth=101))
5 changes: 5 additions & 0 deletions configs/point_rend/pointrend_r50_512x1024_80k_cityscapes.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,5 @@
_base_ = [
'../_base_/models/pointrend_r50.py', '../_base_/datasets/cityscapes.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_80k.py'
]
lr_config = dict(warmup='linear', warmup_iters=200)
32 changes: 32 additions & 0 deletions configs/point_rend/pointrend_r50_512x512_160k_ade20k.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,32 @@
_base_ = [
'../_base_/models/pointrend_r50.py', '../_base_/datasets/ade20k.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_160k.py'
]
norm_cfg = dict(type='SyncBN', requires_grad=True)
model = dict(decode_head=[
dict(
type='FPNHead',
in_channels=[256, 256, 256, 256],
in_index=[0, 1, 2, 3],
feature_strides=[4, 8, 16, 32],
channels=128,
dropout_ratio=-1,
num_classes=150,
norm_cfg=norm_cfg,
align_corners=False,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
dict(
type='PointHead',
in_channels=[256],
in_index=[0],
channels=256,
num_fcs=3,
coarse_pred_each_layer=True,
dropout_ratio=-1,
num_classes=150,
align_corners=False,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0))
])
lr_config = dict(warmup='linear', warmup_iters=200)
16 changes: 16 additions & 0 deletions docs/model_zoo.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -89,6 +89,22 @@ Please refer to [Fast-SCNN](https://github.com/open-mmlab/mmsegmentation/blob/ma

Please refer to [ResNeSt](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/resnest) for details.

### Semantic FPN

Please refer to [Semantic FPN](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/semfpn) for details.

### PointRend

Please refer to [PointRend](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/point_rend) for details.

### EMANet

Please refer to [EMANet](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/emanet) for details.

### DNLNet

Please refer to [DNLNet](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/dnlnet) for details.

### Mixed Precision (FP16) Training

Please refer [Mixed Precision (FP16) Training](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/fp16/README.md) for details.
Expand Down
4 changes: 3 additions & 1 deletion mmseg/models/decode_heads/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,6 +10,7 @@
from .gc_head import GCHead
from .nl_head import NLHead
from .ocr_head import OCRHead
from .point_head import PointHead
from .psa_head import PSAHead
from .psp_head import PSPHead
from .sep_aspp_head import DepthwiseSeparableASPPHead
Expand All @@ -19,5 +20,6 @@
__all__ = [
'FCNHead', 'PSPHead', 'ASPPHead', 'PSAHead', 'NLHead', 'GCHead', 'CCHead',
'UPerHead', 'DepthwiseSeparableASPPHead', 'ANNHead', 'DAHead', 'OCRHead',
'EncHead', 'DepthwiseSeparableFCNHead', 'FPNHead', 'EMAHead', 'DNLHead'
'EncHead', 'DepthwiseSeparableFCNHead', 'FPNHead', 'EMAHead', 'DNLHead',
'PointHead'
]
Loading