Skip to content

Commit e9c41c3

Browse files
MengzhangLIMengzhangLI
authored
[Fix] Fix correct num_classes of HRNet in LoveDA dataset open-mmlab#1136 
1 parent f7ae724 commit e9c41c3

File tree

3 files changed

+19
-18
lines changed

3 files changed

+19
-18
lines changed

‎configs/hrnet/README.md

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -89,6 +89,6 @@ High-resolution representations are essential for position-sensitive vision prob
8989

9090
| Method | Backbone | Crop Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | config | download |
9191
| ---------- | -------- | --------- | ------: | -------- | -------------- | ----: | ------------: | -------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
92-
| 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) | [log](https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr18s_512x512_80k_loveda/fcn_hr18s_512x512_80k_loveda_20211105_180825.log.json) |
93-
| 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) | [log](https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr18_512x512_80k_loveda/fcn_hr18_512x512_80k_loveda_20211105_165542.log.json) |
94-
| 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) | [log](https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr48_512x512_80k_loveda/fcn_hr48_512x512_80k_loveda_20211105_131509.log.json) |
92+
| FCN | HRNetV2p-W18-Small | 512x512 | 80000 | 1.59 | 24.87 | 49.28 | 49.42 | [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_20211210_203228-60a86a7a.pth) | [log](https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr18s_512x512_80k_loveda/fcn_hr18s_512x512_80k_loveda_20211210_203228.log.json) |
93+
| FCN | HRNetV2p-W18 | 512x512 | 80000 | 2.76 | 12.92 | 50.81 | 50.95 | [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_20211210_203952-93d9c3b3.pth) | [log](https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr18_512x512_80k_loveda/fcn_hr18_512x512_80k_loveda_20211210_203952.log.json) |
94+
| FCN | HRNetV2p-W48 | 512x512 | 80000 | 6.20 | 9.61 | 51.42 | 51.64 | [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_20211211_044756-67072f55.pth) | [log](https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr48_512x512_80k_loveda/fcn_hr48_512x512_80k_loveda_20211211_044756.log.json) |

‎configs/hrnet/fcn_hr18_512x512_80k_loveda.py

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -2,3 +2,4 @@
22
'../_base_/models/fcn_hr18.py', '../_base_/datasets/loveda.py',
33
'../_base_/default_runtime.py', '../_base_/schedules/schedule_80k.py'
44
]
5+
model = dict(decode_head=dict(num_classes=7))

‎configs/hrnet/hrnet.yml

Lines changed: 15 additions & 15 deletions
Original file line numberDiff line numberDiff line change
@@ -455,62 +455,62 @@ Models:
455455
crop size: (512,512)
456456
lr schd: 80000
457457
inference time (ms/im):
458-
- value: 33.26
458+
- value: 40.21
459459
hardware: V100
460460
backend: PyTorch
461461
batch size: 1
462462
mode: FP32
463463
resolution: (512,512)
464-
Training Memory (GB): 1.72
464+
Training Memory (GB): 1.59
465465
Results:
466466
- Task: Semantic Segmentation
467467
Dataset: LoveDA
468468
Metrics:
469-
mIoU: 49.3
470-
mIoU(ms+flip): 49.23
469+
mIoU: 49.28
470+
mIoU(ms+flip): 49.42
471471
Config: configs/hrnet/fcn_hr18s_512x512_80k_loveda.py
472-
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr18s_512x512_80k_loveda/fcn_hr18s_512x512_80k_loveda_20211105_180825-41dcc5dc.pth
472+
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr18s_512x512_80k_loveda/fcn_hr18s_512x512_80k_loveda_20211210_203228-60a86a7a.pth
473473
- Name: fcn_hr18_512x512_80k_loveda
474474
In Collection: hrnet
475475
Metadata:
476476
backbone: HRNetV2p-W18
477477
crop size: (512,512)
478478
lr schd: 80000
479479
inference time (ms/im):
480-
- value: 59.63
480+
- value: 77.4
481481
hardware: V100
482482
backend: PyTorch
483483
batch size: 1
484484
mode: FP32
485485
resolution: (512,512)
486-
Training Memory (GB): 2.9
486+
Training Memory (GB): 2.76
487487
Results:
488488
- Task: Semantic Segmentation
489489
Dataset: LoveDA
490490
Metrics:
491-
mIoU: 50.87
492-
mIoU(ms+flip): 51.24
491+
mIoU: 50.81
492+
mIoU(ms+flip): 50.95
493493
Config: configs/hrnet/fcn_hr18_512x512_80k_loveda.py
494-
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr18_512x512_80k_loveda/fcn_hr18_512x512_80k_loveda_20211105_165542-95be4d2b.pth
494+
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr18_512x512_80k_loveda/fcn_hr18_512x512_80k_loveda_20211210_203952-93d9c3b3.pth
495495
- Name: fcn_hr48_512x512_80k_loveda
496496
In Collection: hrnet
497497
Metadata:
498498
backbone: HRNetV2p-W48
499499
crop size: (512,512)
500500
lr schd: 80000
501501
inference time (ms/im):
502-
- value: 110.01
502+
- value: 104.06
503503
hardware: V100
504504
backend: PyTorch
505505
batch size: 1
506506
mode: FP32
507507
resolution: (512,512)
508-
Training Memory (GB): 6.25
508+
Training Memory (GB): 6.2
509509
Results:
510510
- Task: Semantic Segmentation
511511
Dataset: LoveDA
512512
Metrics:
513-
mIoU: 51.04
514-
mIoU(ms+flip): 51.12
513+
mIoU: 51.42
514+
mIoU(ms+flip): 51.64
515515
Config: configs/hrnet/fcn_hr48_512x512_80k_loveda.py
516-
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr48_512x512_80k_loveda/fcn_hr48_512x512_80k_loveda_20211105_131509-f07e47c6.pth
516+
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/hrnet/fcn_hr48_512x512_80k_loveda/fcn_hr48_512x512_80k_loveda_20211211_044756-67072f55.pth

0 commit comments

Comments
 (0)