mirrors/mmsegmentation
1
0
Fork 0
mirror of https://github.com/open-mmlab/mmsegmentation.git synced 2025-06-03 22:03:48 +08:00
Code Issues Projects Releases Wiki Activity

[Feature] Support LoveDA dataset (#1028)

Browse Source 
* update LoveDA dataset api

* revised lint errors in dataset_prepare.md

* revised lint errors in loveda.py

* revised lint errors in loveda.py

* revised lint errors in dataset_prepare.md

* revised lint errors in dataset_prepare.md

* checked with isort and yapf

* checked with isort and yapf

* checked with isort and yapf

* Revert "checked with isort and yapf"

This reverts commit 686a51d9

* Revert "checked with isort and yapf"

This reverts commit b877e121bb2935ceefc503c09675019489829feb.

* Revert "revised lint errors in dataset_prepare.md"

This reverts commit 2289e27c

* Revert "checked with isort and yapf"

This reverts commit 159db2f8

* Revert "checked with isort and yapf"

This reverts commit 159db2f8

* add configs & fix bugs

* update new branch

* upload models&logs and add format-only

* change pretraied model path of HRNet

* fix the errors in dataset_prepare.md

* fix the errors in dataset_prepare.md and configs in loveda.py

* change the description in docs_zh-CN/dataset_prepare.md

* use init_cfg

* fix test converage

* adding pseudo loveda dataset

* adding pseudo loveda dataset

* adding pseudo loveda dataset

* adding pseudo loveda dataset

* adding pseudo loveda dataset

* adding pseudo loveda dataset

* Update docs/dataset_prepare.md

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

* Update docs_zh-CN/dataset_prepare.md

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

* Update docs_zh-CN/dataset_prepare.md

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

* Delete unused lines of unittest and Add docs

* add convert .py file

* add downloading links from zenodo

* move place of LoveDA and Cityscapes in doc

* move place of LoveDA and Cityscapes in doc

Co-authored-by: MengzhangLI <mcmong@pku.edu.cn>
Co-authored-by: Junjun2016 <hejunjun@sjtu.edu.cn>
This commit is contained in:
Kingdrone 2021-11-24 19:41:19 +08:00 committed by GitHub
parent 6070a0e1e7
commit 2bd7f60785
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
32 changed files with 670 additions and 18 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

54
configs/_base_/datasets/loveda.py Normal file
View File

@ -0,0 +1,54 @@
# dataset settings
dataset_type = 'LoveDADataset'
data_root = 'data/loveDA'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
crop_size = (512, 512)
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='LoadAnnotations', reduce_zero_label=True),
dict(type='Resize', img_scale=(2048, 512), ratio_range=(0.5, 2.0)),
dict(type='RandomCrop', crop_size=crop_size, cat_max_ratio=0.75),
dict(type='RandomFlip', prob=0.5),
dict(type='PhotoMetricDistortion'),
dict(type='Normalize', **img_norm_cfg),
dict(type='Pad', size=crop_size, pad_val=0, seg_pad_val=255),
dict(type='DefaultFormatBundle'),
dict(type='Collect', keys=['img', 'gt_semantic_seg']),
]
test_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='MultiScaleFlipAug',
img_scale=(1024, 1024),
# img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75],
flip=False,
transforms=[
dict(type='Resize', keep_ratio=True),
dict(type='RandomFlip'),
dict(type='Normalize', **img_norm_cfg),
dict(type='ImageToTensor', keys=['img']),
dict(type='Collect', keys=['img']),
])
]
data = dict(
samples_per_gpu=4,
workers_per_gpu=4,
train=dict(
type=dataset_type,
data_root=data_root,
img_dir='img_dir/train',
ann_dir='ann_dir/train',
pipeline=train_pipeline),
val=dict(
type=dataset_type,
data_root=data_root,
img_dir='img_dir/val',
ann_dir='ann_dir/val',
pipeline=test_pipeline),
test=dict(
type=dataset_type,
data_root=data_root,
img_dir='img_dir/val',
ann_dir='ann_dir/val',
pipeline=test_pipeline))

8
configs/deeplabv3plus/README.md
View File

@ -85,6 +85,14 @@
| DeepLabV3+ | R-101-D8 | 480x480 | 40000 | - | - | 52.86 | 54.54 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/deeplabv3plus/deeplabv3plus_r101-d8_480x480_40k_pascal_context_59.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/deeplabv3plus/deeplabv3plus_r101-d8_480x480_40k_pascal_context_59/deeplabv3plus_r101-d8_480x480_40k_pascal_context_59_20210416_111233-ed937f15.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/deeplabv3plus/deeplabv3plus_r101-d8_480x480_40k_pascal_context_59/deeplabv3plus_r101-d8_480x480_40k_pascal_context_59-20210416_111233.log.json) |
| DeepLabV3+ | R-101-D8 | 480x480 | 80000 | - | - | 53.2 | 54.67 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/deeplabv3plus/deeplabv3plus_r101-d8_480x480_80k_pascal_context_59.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/deeplabv3plus/deeplabv3plus_r101-d8_480x480_80k_pascal_context_59/deeplabv3plus_r101-d8_480x480_80k_pascal_context_59_20210416_111127-7ca0331d.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/deeplabv3plus/deeplabv3plus_r101-d8_480x480_80k_pascal_context_59/deeplabv3plus_r101-d8_480x480_80k_pascal_context_59-20210416_111127.log.json) |
#### LoveDA
| Method | Backbone | Crop Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | config | download |
| ---------- | -------- | --------- | ------: | -------- | -------------- | ----: | ------------: | -------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| DeepLabV3+ | R-18-D8 | 512x512 | 80000 | 1.93 | 25.57 | 50.28 | 50.47 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/deeplabv3plus/deeplabv3plus_r18-d8_512x512_80k_loveda.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/deeplabv3plus/deeplabv3plus_r18-d8_512x512_80k_loveda/deeplabv3plus_r18-d8_512x512_80k_loveda_20211104_132800-ce0fa0ca.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/deeplabv3plus/deeplabv3plus_r18-d8_512x512_80k_loveda/deeplabv3plus_r18-d8_512x512_80k_loveda_20211104_132800.log.json) |
| DeepLabV3+ | R-50-D8 | 512x512 | 80000 | 7.37 | 6.00 | 50.99 | 50.65 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/deeplabv3plus/deeplabv3plus_r50-d8_512x512_80k_loveda.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/deeplabv3plus/deeplabv3plus_r50-d8_512x512_80k_loveda/deeplabv3plus_r50-d8_512x512_80k_loveda_20211105_080442-f0720392.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/deeplabv3plus/deeplabv3plus_r50-d8_512x512_80k_loveda/deeplabv3plus_r50-d8_512x512_80k_loveda_20211105_080442.log.json) |
| DeepLabV3+ | R-101-D8 | 512x512 | 80000 | 10.84 | 4.33 | 51.47 | 51.32 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/deeplabv3plus/deeplabv3plus_r101-d8_512x512_80k_loveda.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/deeplabv3plus/deeplabv3plus_r101-d8_512x512_80k_loveda/deeplabv3plus_r101-d8_512x512_80k_loveda_20211105_110759-4c1f297e.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/deeplabv3plus/deeplabv3plus_r101-d8_512x512_80k_loveda/deeplabv3plus_r101-d8_512x512_80k_loveda_20211105_110759.log.json) |
Note:
- `FP16` means Mixed Precision (FP16) is adopted in training.

66
configs/deeplabv3plus/deeplabv3plus.yml
View File

@ -599,3 +599,69 @@ Models:
mIoU(ms+flip): 54.67
Config: configs/deeplabv3plus/deeplabv3plus_r101-d8_480x480_80k_pascal_context_59.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/deeplabv3plus/deeplabv3plus_r101-d8_480x480_80k_pascal_context_59/deeplabv3plus_r101-d8_480x480_80k_pascal_context_59_20210416_111127-7ca0331d.pth
- Name: deeplabv3plus_r18-d8_512x512_80k_loveda
In Collection: deeplabv3plus
Metadata:
backbone: R-18-D8
crop size: (512,512)
lr schd: 80000
inference time (ms/im):
- value: 39.11
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
memory (GB): 1.93
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 50.28
mIoU(ms+flip): 50.47
Config: configs/deeplabv3plus/deeplabv3plus_r18-d8_512x512_80k_loveda.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/deeplabv3plus/deeplabv3plus_r18-d8_512x512_80k_loveda/deeplabv3plus_r18-d8_512x512_80k_loveda_20211104_132800-ce0fa0ca.pth
- Name: deeplabv3plus_r50-d8_512x512_80k_loveda
In Collection: deeplabv3plus
Metadata:
backbone: R-50-D8
crop size: (512,512)
lr schd: 80000
inference time (ms/im):
- value: 166.67
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
memory (GB): 7.37
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 50.99
mIoU(ms+flip): 50.65
Config: configs/deeplabv3plus/deeplabv3plus_r50-d8_512x512_80k_loveda.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/deeplabv3plus/deeplabv3plus_r50-d8_512x512_80k_loveda/deeplabv3plus_r50-d8_512x512_80k_loveda_20211105_080442-f0720392.pth
- Name: deeplabv3plus_r101-d8_512x512_80k_loveda
In Collection: deeplabv3plus
Metadata:
backbone: R-101-D8
crop size: (512,512)
lr schd: 80000
inference time (ms/im):
- value: 230.95
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
memory (GB): 10.84
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 51.47
mIoU(ms+flip): 51.32
Config: configs/deeplabv3plus/deeplabv3plus_r101-d8_512x512_80k_loveda.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/deeplabv3plus/deeplabv3plus_r101-d8_512x512_80k_loveda/deeplabv3plus_r101-d8_512x512_80k_loveda_20211105_110759-4c1f297e.pth

6
configs/deeplabv3plus/deeplabv3plus_r101-d8_512x512_80k_loveda.py Normal file
View File

@ -0,0 +1,6 @@
_base_ = './deeplabv3plus_r50-d8_512x512_80k_loveda.py'
model = dict(
backbone=dict(
depth=101,
init_cfg=dict(
type='Pretrained', checkpoint='open-mmlab://resnet101_v1c')))

13
configs/deeplabv3plus/deeplabv3plus_r18-d8_512x512_80k_loveda.py Normal file
View File

@ -0,0 +1,13 @@
_base_ = './deeplabv3plus_r50-d8_512x512_80k_loveda.py'
model = dict(
backbone=dict(
depth=18,
init_cfg=dict(
type='Pretrained', checkpoint='open-mmlab://resnet18_v1c')),
decode_head=dict(
c1_in_channels=64,
c1_channels=12,
in_channels=512,
channels=128,
),
auxiliary_head=dict(in_channels=256, channels=64))

6
configs/deeplabv3plus/deeplabv3plus_r50-d8_512x512_80k_loveda.py Normal file
View File

@ -0,0 +1,6 @@
_base_ = [
'../_base_/models/deeplabv3plus_r50-d8.py', '../_base_/datasets/loveda.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_80k.py'
]
model = dict(
decode_head=dict(num_classes=7), auxiliary_head=dict(num_classes=7))

8
configs/hrnet/README.md
View File

@ -73,3 +73,11 @@
| ------ | ------------ | --------- | ------: | -------- | -------------- | ----: | ------------: | ----------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| FCN | HRNetV2p-W48 | 480x480 | 40000 | - | - | 50.33 | 52.83 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/hrnet/fcn_hr48_480x480_40k_pascal_context_59.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr48_480x480_40k_pascal_context_59/fcn_hr48_480x480_40k_pascal_context_59_20210410_122738-b808b8b2.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr48_480x480_40k_pascal_context_59/fcn_hr48_480x480_40k_pascal_context_59-20210410_122738.log.json) |
| FCN | HRNetV2p-W48 | 480x480 | 80000 | - | - | 51.12 | 53.56 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/hrnet/fcn_hr48_480x480_80k_pascal_context_59.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr48_480x480_80k_pascal_context_59/fcn_hr48_480x480_80k_pascal_context_59_20210411_003240-3ae7081e.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr48_480x480_80k_pascal_context_59/fcn_hr48_480x480_80k_pascal_context_59-20210411_003240.log.json) |
#### LoveDA
| Method | Backbone | Crop Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | config | download |
| ---------- | -------- | --------- | ------: | -------- | -------------- | ----: | ------------: | -------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| FCN | HRNetV2p-W18-Small | 512x512 | 80000 | 1.72 | 30.07 | 49.3 | 49.23 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/hrnet/fcn_hr18s_512x512_80k_loveda.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr18s_512x512_80k_loveda/fcn_hr18s_512x512_80k_loveda_20211105_180825-41dcc5dc.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr18s_512x512_80k_loveda/fcn_hr18s_512x512_80k_loveda_20211105_180825.log.json) |
| FCN | HRNetV2p-W18 | 512x512 | 80000 | 2.90 | 16.77 | 50.87 | 51.24 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/hrnet/fcn_hr18_512x512_80k_loveda.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr18_512x512_80k_loveda/fcn_hr18_512x512_80k_loveda_20211105_165542-95be4d2b.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr18_512x512_80k_loveda/fcn_hr18_512x512_80k_loveda_20211105_165542.log.json) |
| FCN | HRNetV2p-W48 | 512x512 | 80000 | 6.25 | 9.09 | 51.04 | 51.12 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/hrnet/fcn_hr48_512x512_80k_loveda.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr48_512x512_80k_loveda/fcn_hr48_512x512_80k_loveda_20211105_131509-f07e47c6.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr48_512x512_80k_loveda/fcn_hr48_512x512_80k_loveda_20211105_131509.log.json) |

4
configs/hrnet/fcn_hr18_512x512_80k_loveda.py Normal file
View File

@ -0,0 +1,4 @@
_base_ = [
'../_base_/models/fcn_hr18.py', '../_base_/datasets/loveda.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_80k.py'
]

11
configs/hrnet/fcn_hr18s_512x512_80k_loveda.py Normal file
View File

@ -0,0 +1,11 @@
_base_ = './fcn_hr18_512x512_80k_loveda.py'
model = dict(
backbone=dict(
init_cfg=dict(
type='Pretrained',
checkpoint='open-mmlab://msra/hrnetv2_w18_small'),
extra=dict(
stage1=dict(num_blocks=(2, )),
stage2=dict(num_blocks=(2, 2)),
stage3=dict(num_modules=3, num_blocks=(2, 2, 2)),
stage4=dict(num_modules=2, num_blocks=(2, 2, 2, 2)))))

11
configs/hrnet/fcn_hr48_512x512_80k_loveda.py Normal file
View File

@ -0,0 +1,11 @@
_base_ = './fcn_hr18_512x512_80k_loveda.py'
model = dict(
backbone=dict(
init_cfg=dict(
type='Pretrained', checkpoint='open-mmlab://msra/hrnetv2_w48'),
extra=dict(
stage2=dict(num_channels=(48, 96)),
stage3=dict(num_channels=(48, 96, 192)),
stage4=dict(num_channels=(48, 96, 192, 384)))),
decode_head=dict(
in_channels=[48, 96, 192, 384], channels=sum([48, 96, 192, 384])))

66
configs/hrnet/hrnet.yml
View File

@ -447,3 +447,69 @@ Models:
mIoU(ms+flip): 53.56
Config: configs/hrnet/fcn_hr48_480x480_80k_pascal_context_59.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr48_480x480_80k_pascal_context_59/fcn_hr48_480x480_80k_pascal_context_59_20210411_003240-3ae7081e.pth
- Name: fcn_hr18s_512x512_80k_loveda
In Collection: hrnet
Metadata:
backbone: HRNetV2p-W18-Small
crop size: (512,512)
lr schd: 80000
inference time (ms/im):
- value: 33.26
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
memory (GB): 1.72
Results:
- Task: Semantic Segmentation
Dataset: Pascal Context 59
Metrics:
mIoU: 49.3
mIoU(ms+flip): 49.23
Config: configs/hrnet/fcn_hr18s_512x512_80k_loveda.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr18s_512x512_80k_loveda/fcn_hr18s_512x512_80k_loveda_20211105_180825-41dcc5dc.pth
- Name: fcn_hr18_512x512_80k_loveda
In Collection: hrnet
Metadata:
backbone: HRNetV2p-W18
crop size: (512,512)
lr schd: 80000
inference time (ms/im):
- value: 59.63
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
memory (GB): 2.9
Results:
- Task: Semantic Segmentation
Dataset: Pascal Context 59
Metrics:
mIoU: 50.87
mIoU(ms+flip): 51.24
Config: configs/hrnet/fcn_hr18_512x512_80k_loveda.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr18_512x512_80k_loveda/fcn_hr18_512x512_80k_loveda_20211105_165542-95be4d2b.pth
- Name: fcn_hr48_512x512_80k_loveda
In Collection: hrnet
Metadata:
backbone: HRNetV2p-W48
crop size: (512,512)
lr schd: 80000
inference time (ms/im):
- value: 110.01
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
memory (GB): 6.25
Results:
- Task: Semantic Segmentation
Dataset: Pascal Context 59
Metrics:
mIoU: 51.04
mIoU(ms+flip): 51.12
Config: configs/hrnet/fcn_hr48_512x512_80k_loveda.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr48_512x512_80k_loveda/fcn_hr48_512x512_80k_loveda_20211105_131509-f07e47c6.pth

28
configs/pspnet/README.md
View File

@ -98,21 +98,29 @@ We support evaluation results on these two datasets using models above trained o
| Method | Backbone | Crop Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | config | download |
| --------- | -------- | --------- | ------: | -------- | -------------- | ----: | ------------: | ----------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| PSPNet | R-50-D8 | 512x512 | 20000 | 9.6 | 20.5 | 35.69 | 36.62 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet_r50-d8_512x512_4x4_20k_coco-stuff10k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_4x4_20k_coco-stuff10k/pspnet_r50-d8_512x512_4x4_20k_coco-stuff10k_20210820_203258-b88df27f.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_4x4_20k_coco-stuff10k/pspnet_r50-d8_512x512_4x4_20k_coco-stuff10k_20210820_203258.log.json) |
| PSPNet | R-101-D8 | 512x512 | 20000 | 13.2 | 11.1 | 37.26 | 38.52 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet_r101-d8_512x512_4x4_20k_coco-stuff10k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_4x4_20k_coco-stuff10k/pspnet_r101-d8_512x512_4x4_20k_coco-stuff10k_20210820_232135-76aae482.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_4x4_20k_coco-stuff10k/pspnet_r101-d8_512x512_4x4_20k_coco-stuff10k_20210820_232135.log.json) |
| PSPNet | R-50-D8 | 512x512 | 40000 | - | - | 36.33 | 37.24 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet_r50-d8_512x512_4x4_40k_coco-stuff10k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_4x4_40k_coco-stuff10k/pspnet_r50-d8_512x512_4x4_40k_coco-stuff10k_20210821_030857-92e2902b.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_4x4_40k_coco-stuff10k/pspnet_r50-d8_512x512_4x4_40k_coco-stuff10k_20210821_030857.log.json) |
| PSPNet | R-101-D8 | 512x512 | 40000 | - | - | 37.76 | 38.86 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet_r101-d8_512x512_4x4_40k_coco-stuff10k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_4x4_40k_coco-stuff10k/pspnet_r101-d8_512x512_4x4_40k_coco-stuff10k_20210821_014022-831aec95.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_4x4_40k_coco-stuff10k/pspnet_r101-d8_512x512_4x4_40k_coco-stuff10k_20210821_014022.log.json) |
| PSPNet | R-50-D8 | 512x512 | 20000 | 9.6 | 20.5 | 35.69 | 36.62 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet/pspnet_r50-d8_512x512_4x4_20k_coco-stuff10k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_4x4_20k_coco-stuff10k/pspnet_r50-d8_512x512_4x4_20k_coco-stuff10k_20210820_203258-b88df27f.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_4x4_20k_coco-stuff10k/pspnet_r50-d8_512x512_4x4_20k_coco-stuff10k_20210820_203258.log.json) |
| PSPNet | R-101-D8 | 512x512 | 20000 | 13.2 | 11.1 | 37.26 | 38.52 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet/pspnet_r101-d8_512x512_4x4_20k_coco-stuff10k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_4x4_20k_coco-stuff10k/pspnet_r101-d8_512x512_4x4_20k_coco-stuff10k_20210820_232135-76aae482.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_4x4_20k_coco-stuff10k/pspnet_r101-d8_512x512_4x4_20k_coco-stuff10k_20210820_232135.log.json) |
| PSPNet | R-50-D8 | 512x512 | 40000 | - | - | 36.33 | 37.24 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet/pspnet_r50-d8_512x512_4x4_40k_coco-stuff10k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_4x4_40k_coco-stuff10k/pspnet_r50-d8_512x512_4x4_40k_coco-stuff10k_20210821_030857-92e2902b.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_4x4_40k_coco-stuff10k/pspnet_r50-d8_512x512_4x4_40k_coco-stuff10k_20210821_030857.log.json) |
| PSPNet | R-101-D8 | 512x512 | 40000 | - | - | 37.76 | 38.86 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet/pspnet_r101-d8_512x512_4x4_40k_coco-stuff10k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_4x4_40k_coco-stuff10k/pspnet_r101-d8_512x512_4x4_40k_coco-stuff10k_20210821_014022-831aec95.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_4x4_40k_coco-stuff10k/pspnet_r101-d8_512x512_4x4_40k_coco-stuff10k_20210821_014022.log.json) |
### COCO-Stuff 164k
| Method | Backbone | Crop Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | config | download |
| --------- | -------- | --------- | ------: | -------- | -------------- | ----: | ------------: | ----------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| PSPNet | R-50-D8 | 512x512 | 80000 | 9.6 | 20.5 | 38.80 | 39.19 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet_r50-d8_512x512_4x4_80k_coco-stuff164k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_4x4_80k_coco-stuff164k/pspnet_r50-d8_512x512_4x4_80k_coco-stuff164k_20210707_152034-0e41b2db.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_4x4_80k_coco-stuff164k/pspnet_r50-d8_512x512_4x4_80k_coco-stuff164k_20210707_152034.log.json) |
| PSPNet | R-101-D8 | 512x512 | 80000 | 13.2 | 11.1 | 40.34 | 40.79 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet_r101-d8_512x512_4x4_80k_coco-stuff164k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_4x4_80k_coco-stuff164k/pspnet_r101-d8_512x512_4x4_80k_coco-stuff164k_20210707_152034-7eb41789.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_4x4_80k_coco-stuff164k/pspnet_r101-d8_512x512_4x4_80k_coco-stuff164k_20210707_152034.log.json) |
| PSPNet | R-50-D8 | 512x512 | 160000 | - | - | 39.64 | 39.97 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet_r50-d8_512x512_4x4_160k_coco-stuff164k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_4x4_160k_coco-stuff164k/pspnet_r50-d8_512x512_4x4_160k_coco-stuff164k_20210707_152004-51276a57.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_4x4_160k_coco-stuff164k/pspnet_r50-d8_512x512_4x4_160k_coco-stuff164k_20210707_152004.log.json) |
| PSPNet | R-101-D8 | 512x512 | 160000 | - | - | 41.28 | 41.66 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet_r101-d8_512x512_4x4_160k_coco-stuff164k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_4x4_160k_coco-stuff164k/pspnet_r101-d8_512x512_4x4_160k_coco-stuff164k_20210707_152004-4af9621b.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_4x4_160k_coco-stuff164k/pspnet_r101-d8_512x512_4x4_160k_coco-stuff164k_20210707_152004.log.json) |
| PSPNet | R-50-D8 | 512x512 | 320000 | - | - | 40.53 | 40.75 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet_r50-d8_512x512_4x4_320k_coco-stuff164k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_4x4_320k_coco-stuff164k/pspnet_r50-d8_512x512_4x4_320k_coco-stuff164k_20210707_152004-be9610cc.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_4x4_320k_coco-stuff164k/pspnet_r50-d8_512x512_4x4_320k_coco-stuff164k_20210707_152004.log.json) |
| PSPNet | R-101-D8 | 512x512 | 320000 | - | - | 41.95 | 42.42 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet_r101-d8_512x512_4x4_320k_coco-stuff164k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_4x4_320k_coco-stuff164k/pspnet_r101-d8_512x512_4x4_320k_coco-stuff164k_20210707_152004-72220c60.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_4x4_320k_coco-stuff164k/pspnet_r101-d8_512x512_4x4_320k_coco-stuff164k_20210707_152004.log.json) |
| PSPNet | R-50-D8 | 512x512 | 80000 | 9.6 | 20.5 | 38.80 | 39.19 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet/pspnet_r50-d8_512x512_4x4_80k_coco-stuff164k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_4x4_80k_coco-stuff164k/pspnet_r50-d8_512x512_4x4_80k_coco-stuff164k_20210707_152034-0e41b2db.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_4x4_80k_coco-stuff164k/pspnet_r50-d8_512x512_4x4_80k_coco-stuff164k_20210707_152034.log.json) |
| PSPNet | R-101-D8 | 512x512 | 80000 | 13.2 | 11.1 | 40.34 | 40.79 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet/pspnet_r101-d8_512x512_4x4_80k_coco-stuff164k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_4x4_80k_coco-stuff164k/pspnet_r101-d8_512x512_4x4_80k_coco-stuff164k_20210707_152034-7eb41789.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_4x4_80k_coco-stuff164k/pspnet_r101-d8_512x512_4x4_80k_coco-stuff164k_20210707_152034.log.json) |
| PSPNet | R-50-D8 | 512x512 | 160000 | - | - | 39.64 | 39.97 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet/pspnet_r50-d8_512x512_4x4_160k_coco-stuff164k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_4x4_160k_coco-stuff164k/pspnet_r50-d8_512x512_4x4_160k_coco-stuff164k_20210707_152004-51276a57.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_4x4_160k_coco-stuff164k/pspnet_r50-d8_512x512_4x4_160k_coco-stuff164k_20210707_152004.log.json) |
| PSPNet | R-101-D8 | 512x512 | 160000 | - | - | 41.28 | 41.66 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet/pspnet_r101-d8_512x512_4x4_160k_coco-stuff164k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_4x4_160k_coco-stuff164k/pspnet_r101-d8_512x512_4x4_160k_coco-stuff164k_20210707_152004-4af9621b.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_4x4_160k_coco-stuff164k/pspnet_r101-d8_512x512_4x4_160k_coco-stuff164k_20210707_152004.log.json) |
| PSPNet | R-50-D8 | 512x512 | 320000 | - | - | 40.53 | 40.75 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet/pspnet_r50-d8_512x512_4x4_320k_coco-stuff164k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_4x4_320k_coco-stuff164k/pspnet_r50-d8_512x512_4x4_320k_coco-stuff164k_20210707_152004-be9610cc.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_4x4_320k_coco-stuff164k/pspnet_r50-d8_512x512_4x4_320k_coco-stuff164k_20210707_152004.log.json) |
| PSPNet | R-101-D8 | 512x512 | 320000 | - | - | 41.95 | 42.42 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet/pspnet_r101-d8_512x512_4x4_320k_coco-stuff164k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_4x4_320k_coco-stuff164k/pspnet_r101-d8_512x512_4x4_320k_coco-stuff164k_20210707_152004-72220c60.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_4x4_320k_coco-stuff164k/pspnet_r101-d8_512x512_4x4_320k_coco-stuff164k_20210707_152004.log.json) |
#### LoveDA
| Method | Backbone | Crop Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | config | download |
| ---------- | -------- | --------- | ------: | -------- | -------------- | ----: | ------------: | -------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| PSPNet | R-18-D8 | 512x512 | 80000 | 1.45 | 26.87 | 48.62 | 47.57 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet/pspnet_r18-d8_512x512_80k_loveda.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r18-d8_512x512_80k_loveda/pspnet_r18-d8_512x512_80k_loveda_20211105_052100-b97697f1.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r18-d8_512x512_80k_loveda/pspnet_r18-d8_512x512_80k_loveda_20211105_052100.log.json) |
| PSPNet | R-50-D8 | 512x512 | 80000 | 6.14 | 6.60 | 50.46 | 50.19 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet/pspnet_r50-d8_512x512_80k_loveda.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_80k_loveda/pspnet_r50-d8_512x512_80k_loveda_20211104_155728-88610f9f.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_80k_loveda/pspnet_r50-d8_512x512_80k_loveda_20211104_155728.log.json) |
| PSPNet | R-101-D8 | 512x512 | 80000 | 9.61 | 4.58 | 51.86 | 51.34 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/pspnet/pspnet_r101-d8_512x512_80k_loveda.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_80k_loveda/pspnet_r101-d8_512x512_80k_loveda_20211104_153212-1c06c6a8.pth) &#124; [log](https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_80k_loveda/pspnet_r101-d8_512x512_80k_loveda_20211104_153212.log.json) |
Note:

66
configs/pspnet/pspnet.yml
View File

@ -741,3 +741,69 @@ Models:
mIoU(ms+flip): 42.42
Config: configs/pspnet/pspnet_r101-d8_512x512_4x4_320k_coco-stuff164k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_4x4_320k_coco-stuff164k/pspnet_r101-d8_512x512_4x4_320k_coco-stuff164k_20210707_152004-72220c60.pth
- Name: pspnet_r18-d8_512x512_80k_loveda
In Collection: pspnet
Metadata:
backbone: R-18-D8
crop size: (512,512)
lr schd: 80000
inference time (ms/im):
- value: 37.22
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
memory (GB): 1.45
Results:
- Task: Semantic Segmentation
Dataset: COCO-Stuff 164k
Metrics:
mIoU: 48.62
mIoU(ms+flip): 47.57
Config: configs/pspnet/pspnet_r18-d8_512x512_80k_loveda.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r18-d8_512x512_80k_loveda/pspnet_r18-d8_512x512_80k_loveda_20211105_052100-b97697f1.pth
- Name: pspnet_r50-d8_512x512_80k_loveda
In Collection: pspnet
Metadata:
backbone: R-50-D8
crop size: (512,512)
lr schd: 80000
inference time (ms/im):
- value: 151.52
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
memory (GB): 6.14
Results:
- Task: Semantic Segmentation
Dataset: COCO-Stuff 164k
Metrics:
mIoU: 50.46
mIoU(ms+flip): 50.19
Config: configs/pspnet/pspnet_r50-d8_512x512_80k_loveda.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r50-d8_512x512_80k_loveda/pspnet_r50-d8_512x512_80k_loveda_20211104_155728-88610f9f.pth
- Name: pspnet_r101-d8_512x512_80k_loveda
In Collection: pspnet
Metadata:
backbone: R-101-D8
crop size: (512,512)
lr schd: 80000
inference time (ms/im):
- value: 218.34
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
memory (GB): 9.61
Results:
- Task: Semantic Segmentation
Dataset: COCO-Stuff 164k
Metrics:
mIoU: 51.86
mIoU(ms+flip): 51.34
Config: configs/pspnet/pspnet_r101-d8_512x512_80k_loveda.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/pspnet/pspnet_r101-d8_512x512_80k_loveda/pspnet_r101-d8_512x512_80k_loveda_20211104_153212-1c06c6a8.pth

6
configs/pspnet/pspnet_r101-d8_512x512_80k_loveda.py Normal file
View File

@ -0,0 +1,6 @@
_base_ = './pspnet_r50-d8_512x512_80k_loveda.py'
model = dict(
backbone=dict(
depth=101,
init_cfg=dict(
type='Pretrained', checkpoint='open-mmlab://resnet101_v1c')))

11
configs/pspnet/pspnet_r18-d8_512x512_80k_loveda.py Normal file
View File

@ -0,0 +1,11 @@
_base_ = './pspnet_r50-d8_512x512_80k_loveda.py'
model = dict(
backbone=dict(
depth=18,
init_cfg=dict(
type='Pretrained', checkpoint='open-mmlab://resnet18_v1c')),
decode_head=dict(
in_channels=512,
channels=128,
),
auxiliary_head=dict(in_channels=256, channels=64))

6
configs/pspnet/pspnet_r50-d8_512x512_80k_loveda.py Normal file
View File

@ -0,0 +1,6 @@
_base_ = [
'../_base_/models/pspnet_r50-d8.py', '../_base_/datasets/loveda.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_80k.py'
]
model = dict(
decode_head=dict(num_classes=7), auxiliary_head=dict(num_classes=7))

33
docs/dataset_prepare.md
View File

@ -108,6 +108,14 @@ mmsegmentation
| | └── leftImg8bit
| | | └── test
| | | └── night
│ ├── loveDA
│ │ ├── img_dir
│ │ │ ├── train
│ │ │ ├── val
│ │ │ ├── test
│ │ ├── ann_dir
│ │ │ ├── train
│ │ │ ├── val
```
### Cityscapes
@ -253,3 +261,28 @@ Since we only support test models on this dataset, you may only download [the va
### Nighttime Driving
Since we only support test models on this dataset, you may only download [the test set](http://data.vision.ee.ethz.ch/daid/NighttimeDriving/NighttimeDrivingTest.zip).
### LoveDA
The data could be downloaded from Google Drive [here](https://drive.google.com/drive/folders/1ibYV0qwn4yuuh068Rnc-w4tPi0U0c-ti?usp=sharing).
Or it can be downloaded from [zenodo](https://zenodo.org/record/5706578#.YZvN7SYRXdF), you should run the following command:
```shell
# Download Train.zip
wget https://zenodo.org/record/5706578/files/Train.zip
# Download Val.zip
wget https://zenodo.org/record/5706578/files/Val.zip
# Download Test.zip
wget https://zenodo.org/record/5706578/files/Test.zip
```
For LoveDA dataset, please run the following command to download and re-organize the dataset.
```shell
python tools/convert_datasets/loveda.py /path/to/loveDA
```
Using trained model to predict test set of LoveDA and submit it to server can be found [here](https://github.com/open-mmlab/mmsegmentation/blob/master/docs/inference.md).
More details about LoveDA can be found [here](https://github.com/Junjue-Wang/LoveDA).

22
docs/inference.md
View File

@ -101,3 +101,25 @@ Assume that you have already downloaded the checkpoints to the directory `checkp
```
Using ```pmap``` to view CPU memory footprint, it used 2.25GB CPU memory with ```efficient_test=True``` and 11.06GB CPU memory with ```efficient_test=False``` . This optional parameter can save a lot of memory. (After mmseg v0.17, efficient_test has not effect and we use a progressive mode to evaluation and format results efficiently by default.)
7. Test PSPNet on LoveDA test split with 1 GPU, and generate the png files to be submit to the official evaluation server.
First, add following to config file `configs/pspnet/pspnet_r50-d8_512x512_80k_loveda.py`,
```python
data = dict(
test=dict(
img_dir='img_dir/test',
ann_dir='ann_dir/test'))
```
Then run test.
```shell
python ./tools/test.py configs/pspnet/pspnet_r50-d8_512x512_80k_loveda.py \
checkpoints/pspnet_r50-d8_512x512_80k_loveda_20211104_155728-88610f9f.pth \
--format-only --eval-options "imgfile_prefix=./pspnet_test_results"
```
You will get png files under `./pspnet_test_results` directory.
You may run `zip -r -j Results.zip pspnet_test_results/` and submit the zip file to [evaluation server](https://competitions.codalab.org/competitions/35865#participate-submit_results).

37
docs_zh-CN/dataset_prepare.md
View File

@ -89,6 +89,14 @@ mmsegmentation
| | └── leftImg8bit
| | | └── test
| | | └── night
│ ├── loveDA
│ │ ├── img_dir
│ │ │ ├── train
│ │ │ ├── val
│ │ │ ├── test
│ │ ├── ann_dir
│ │ │ ├── train
│ │ │ ├── val
```
### Cityscapes
@ -190,8 +198,33 @@ python tools/convert_datasets/stare.py /path/to/stare-images.tar /path/to/labels
### Dark Zurich
因为我们只支持在此数据集上测试模型，所以您只需下载[验证集](https://data.vision.ee.ethz.ch/csakarid/shared/GCMA_UIoU/Dark_Zurich_val_anon.zip)。
因为我们只支持在此数据集上测试模型，所以您只需下载[验证集](https://data.vision.ee.ethz.ch/csakarid/shared/GCMA_UIoU/Dark_Zurich_val_anon.zip) 。
### Nighttime Driving
因为我们只支持在此数据集上测试模型，所以您只需下载[测试集](http://data.vision.ee.ethz.ch/daid/NighttimeDriving/NighttimeDrivingTest.zip)。
因为我们只支持在此数据集上测试模型，所以您只需下载[测试集](http://data.vision.ee.ethz.ch/daid/NighttimeDriving/NighttimeDrivingTest.zip) 。
### LoveDA
可以从 Google Drive 里下载 [LoveDA数据集](https://drive.google.com/drive/folders/1ibYV0qwn4yuuh068Rnc-w4tPi0U0c-ti?usp=sharing) 。
或者它还可以从 [zenodo](https://zenodo.org/record/5706578#.YZvN7SYRXdF) 下载, 您需要运行如下命令:
```shell
# Download Train.zip
wget https://zenodo.org/record/5706578/files/Train.zip
# Download Val.zip
wget https://zenodo.org/record/5706578/files/Val.zip
# Download Test.zip
wget https://zenodo.org/record/5706578/files/Test.zip
```
对于 LoveDA 数据集，请运行以下命令下载并重新组织数据集
```shell
python tools/convert_datasets/loveda.py /path/to/loveDA
```
请参照 [这里](https://github.com/open-mmlab/mmsegmentation/blob/master/docs_zh-CN/inference.md) 来使用训练好的模型去预测 LoveDA 测试集并且提交到官网。
关于 LoveDA 的更多细节可以在[这里](https://github.com/Junjue-Wang/LoveDA) 找到。

24
docs_zh-CN/inference.md
View File

@ -84,7 +84,7 @@ python tools/test.py ${配置文件} ${检查点文件} [--out ${结果文件}]
```
您会在文件夹 `./pspnet_test_results` 里得到生成的 png 文件。
您也许可以运行 `zip -r results.zip pspnet_test_results/` 并提交 zip 文件给 [evaluation server](https://www.cityscapes-dataset.com/submit/)。
您也许可以运行 `zip -r results.zip pspnet_test_results/` 并提交 zip 文件给 [evaluation server](https://www.cityscapes-dataset.com/submit/) 。
6. 在 Cityscapes 数据集上使用 CPU 高效内存选项来测试 DeeplabV3+ `mIoU` 指标 (没有保存测试结果)
@ -97,3 +97,25 @@ python tools/test.py ${配置文件} ${检查点文件} [--out ${结果文件}]
```
使用 ```pmap``` 可查看 CPU 内存情况, ```efficient_test=True``` 会使用约 2.25GB 的 CPU 内存， ```efficient_test=False``` 会使用约 11.06GB 的 CPU 内存。 这个可选参数可以节约很多 CPU 内存。（MMseg v0.17 之后, `efficient_test` 参数将不再生效, 我们使用了一种渐近的方式来更加有效快速地评估和保存结果。）
7. 在 LoveDA 数据集上1卡 GPU 测试 PSPNet， 并生成 png 文件以便提交给官方评估服务器
首先，在配置文件里添加内容： `configs/pspnet/pspnet_r50-d8_512x512_80k_loveda.py`,
```python
data = dict(
test=dict(
img_dir='img_dir/test',
ann_dir='ann_dir/test'))
```
随后，进行测试。
```shell
python ./tools/test.py configs/pspnet/pspnet_r50-d8_512x512_80k_loveda.py \
checkpoints/pspnet_r50-d8_512x512_80k_loveda_20211104_155728-88610f9f.pth \
--format-only --eval-options "imgfile_prefix=./pspnet_test_results"
```
您会在文件夹 `./pspnet_test_results` 里得到生成的 png 文件。
您也许可以运行 `zip -r -j Results.zip pspnet_test_results/` 并提交 zip 文件给 [evaluation server](https://competitions.codalab.org/competitions/35865#participate-submit_results) 。

3
mmseg/datasets/__init__.py
View File

@ -9,6 +9,7 @@ from .dark_zurich import DarkZurichDataset
from .dataset_wrappers import ConcatDataset, RepeatDataset
from .drive import DRIVEDataset
from .hrf import HRFDataset
from .loveda import LoveDADataset
from .night_driving import NightDrivingDataset
from .pascal_context import PascalContextDataset, PascalContextDataset59
from .stare import STAREDataset
@ -20,5 +21,5 @@ __all__ = [
'PascalVOCDataset', 'ADE20KDataset', 'PascalContextDataset',
'PascalContextDataset59', 'ChaseDB1Dataset', 'DRIVEDataset', 'HRFDataset',
'STAREDataset', 'DarkZurichDataset', 'NightDrivingDataset',
'COCOStuffDataset'
'COCOStuffDataset', 'LoveDADataset'
]

2
mmseg/datasets/ade.py
View File

@ -94,7 +94,7 @@ class ADE20KDataset(CustomDataset):
"""Write the segmentation results to images.
Args:
results (list[list | tuple | ndarray]): Testing results of the
results (list[ndarray]): Testing results of the
dataset.
imgfile_prefix (str): The filename prefix of the png files.
If the prefix is "somepath/xxx",

2
mmseg/datasets/cityscapes.py
View File

@ -52,7 +52,7 @@ class CityscapesDataset(CustomDataset):
"""Write the segmentation results to images.
Args:
results (list[list | tuple | ndarray]): Testing results of the
results (list[ndarray]): Testing results of the
dataset.
imgfile_prefix (str): The filename prefix of the png files.
If the prefix is "somepath/xxx",

92
mmseg/datasets/loveda.py Normal file
View File

@ -0,0 +1,92 @@
# Copyright (c) OpenMMLab. All rights reserved.
import os.path as osp
import mmcv
import numpy as np
from PIL import Image
from .builder import DATASETS
from .custom import CustomDataset
@DATASETS.register_module()
class LoveDADataset(CustomDataset):
"""LoveDA dataset.
In segmentation map annotation for LoveDA, 0 is the ignore index.
``reduce_zero_label`` should be set to True. The ``img_suffix`` and
``seg_map_suffix`` are both fixed to '.png'.
"""
CLASSES = ('background', 'building', 'road', 'water', 'barren', 'forest',
'agricultural')
PALETTE = [[255, 255, 255], [255, 0, 0], [255, 255, 0], [0, 0, 255],
[159, 129, 183], [0, 255, 0], [255, 195, 128]]
def __init__(self, **kwargs):
super(LoveDADataset, self).__init__(
img_suffix='.png',
seg_map_suffix='.png',
reduce_zero_label=True,
**kwargs)
def results2img(self, results, imgfile_prefix, indices=None):
"""Write the segmentation results to images.
Args:
results (list[ndarray]): Testing results of the
dataset.
imgfile_prefix (str): The filename prefix of the png files.
If the prefix is "somepath/xxx",
the png files will be named "somepath/xxx.png".
indices (list[int], optional): Indices of input results, if not
set, all the indices of the dataset will be used.
Default: None.
Returns:
list[str: str]: result txt files which contains corresponding
semantic segmentation images.
"""
mmcv.mkdir_or_exist(imgfile_prefix)
result_files = []
for result, idx in zip(results, indices):
filename = self.img_infos[idx]['filename']
basename = osp.splitext(osp.basename(filename))[0]
png_filename = osp.join(imgfile_prefix, f'{basename}.png')
# The index range of official requirement is from 0 to 6.
output = Image.fromarray(result.astype(np.uint8))
output.save(png_filename)
result_files.append(png_filename)
return result_files
def format_results(self, results, imgfile_prefix, indices=None):
"""Format the results into dir (standard format for LoveDA evaluation).
Args:
results (list): Testing results of the dataset.
imgfile_prefix (str): The prefix of images files. It
includes the file path and the prefix of filename, e.g.,
"a/b/prefix".
indices (list[int], optional): Indices of input results,
if not set, all the indices of the dataset will be used.
Default: None.
Returns:
tuple: (result_files, tmp_dir), result_files is a list containing
the image paths, tmp_dir is the temporal directory created
for saving json/png files when img_prefix is not specified.
"""
if indices is None:
indices = list(range(len(self)))
assert isinstance(results, list), 'results must be a list.'
assert isinstance(indices, list), 'indices must be a list.'
result_files = self.results2img(results, imgfile_prefix, indices)
return result_files

BIN
tests/data/pseudo_loveda_dataset/ann_dir/0.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 6.0 KiB

BIN
tests/data/pseudo_loveda_dataset/ann_dir/1.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 13 KiB

BIN
tests/data/pseudo_loveda_dataset/ann_dir/2.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 5.4 KiB

BIN
tests/data/pseudo_loveda_dataset/img_dir/0.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 1.7 MiB

BIN
tests/data/pseudo_loveda_dataset/img_dir/1.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 1.7 MiB

BIN
tests/data/pseudo_loveda_dataset/img_dir/2.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 2.0 MiB

30
tests/test_data/test_dataset.py
View File

@ -13,8 +13,8 @@ from PIL import Image
from mmseg.core.evaluation import get_classes, get_palette
from mmseg.datasets import (DATASETS, ADE20KDataset, CityscapesDataset,
ConcatDataset, CustomDataset, PascalVOCDataset,
RepeatDataset, build_dataset)
ConcatDataset, CustomDataset, LoveDADataset,
PascalVOCDataset, RepeatDataset, build_dataset)
def test_classes():
@ -622,6 +622,32 @@ def test_concat_cityscapes(separate_eval):
separate_eval=separate_eval)
def test_loveda():
test_dataset = LoveDADataset(
pipeline=[],
img_dir=osp.join(
osp.dirname(__file__), '../data/pseudo_loveda_dataset/img_dir'),
ann_dir=osp.join(
osp.dirname(__file__), '../data/pseudo_loveda_dataset/ann_dir'))
assert len(test_dataset) == 3
gt_seg_maps = list(test_dataset.get_gt_seg_maps())
# Test format_results
pseudo_results = []
for idx in range(len(test_dataset)):
h, w = gt_seg_maps[idx].shape
pseudo_results.append(np.random.randint(low=0, high=7, size=(h, w)))
file_paths = test_dataset.format_results(pseudo_results, '.format_loveda')
assert len(file_paths) == len(test_dataset)
# Test loveda evaluate
test_dataset.evaluate(
pseudo_results, metric='mIoU', imgfile_prefix='.format_loveda')
shutil.rmtree('.format_loveda')
@patch('mmseg.datasets.CustomDataset.load_annotations', MagicMock)
@patch('mmseg.datasets.CustomDataset.__getitem__',
MagicMock(side_effect=lambda idx: idx))

73
tools/convert_datasets/loveda.py Normal file
View File

@ -0,0 +1,73 @@
# Copyright (c) OpenMMLab. All rights reserved.
import argparse
import os
import os.path as osp
import shutil
import tempfile
import zipfile
import mmcv
def parse_args():
parser = argparse.ArgumentParser(
description='Convert LoveDA dataset to mmsegmentation format')
parser.add_argument('dataset_path', help='LoveDA folder path')
parser.add_argument('--tmp_dir', help='path of the temporary directory')
parser.add_argument('-o', '--out_dir', help='output path')
args = parser.parse_args()
return args
def main():
args = parse_args()
dataset_path = args.dataset_path
if args.out_dir is None:
out_dir = osp.join('data', 'loveDA')
else:
out_dir = args.out_dir
print('Making directories...')
mmcv.mkdir_or_exist(out_dir)
mmcv.mkdir_or_exist(osp.join(out_dir, 'img_dir'))
mmcv.mkdir_or_exist(osp.join(out_dir, 'img_dir', 'train'))
mmcv.mkdir_or_exist(osp.join(out_dir, 'img_dir', 'val'))
mmcv.mkdir_or_exist(osp.join(out_dir, 'img_dir', 'test'))
mmcv.mkdir_or_exist(osp.join(out_dir, 'ann_dir'))
mmcv.mkdir_or_exist(osp.join(out_dir, 'ann_dir', 'train'))
mmcv.mkdir_or_exist(osp.join(out_dir, 'ann_dir', 'val'))
assert 'Train.zip' in os.listdir(dataset_path), \
'Train.zip is not in {}'.format(dataset_path)
assert 'Val.zip' in os.listdir(dataset_path), \
'Val.zip is not in {}'.format(dataset_path)
assert 'Test.zip' in os.listdir(dataset_path), \
'Test.zip is not in {}'.format(dataset_path)
with tempfile.TemporaryDirectory(dir=args.tmp_dir) as tmp_dir:
for dataset in ['Train', 'Val', 'Test']:
zip_file = zipfile.ZipFile(
os.path.join(dataset_path, dataset + '.zip'))
zip_file.extractall(tmp_dir)
data_type = dataset.lower()
for location in ['Rural', 'Urban']:
for image_type in ['images_png', 'masks_png']:
if image_type == 'images_png':
dst = osp.join(out_dir, 'img_dir', data_type)
else:
dst = osp.join(out_dir, 'ann_dir', data_type)
if dataset == 'Test' and image_type == 'masks_png':
continue
else:
src_dir = osp.join(tmp_dir, dataset, location,
image_type)
src_lst = os.listdir(src_dir)
for file in src_lst:
shutil.move(osp.join(src_dir, file), dst)
print('Removing the temporary files...')
print('Done!')
if __name__ == '__main__':
main()