Semantic Segmentation on ImageNet-S

Metric: mIoU (val) (higher is better)

LeaderboardDataset

Loading chart...

Results

Sort:

#	Model↕	mIoU (val)▼	Extra Data	Paper	Date↕	Code
1	TEC (ViT-B/16, 224x224, SSL+FT, mmseg)	63.2	No	Towards Sustainable Self-supervised Learning	2022-10-20	Code
2	SERE (ViT-B/16, 100ep, 224x224, SSL+FT)	63	No	SERE: Exploring Feature Self-relation for Self-s...	2022-06-10	Code
3	TEC (ViT-B/16, 224x224, SSL+FT)	62	No	Towards Sustainable Self-supervised Learning	2022-10-20	Code
4	MAE (ViT-B/16, 224x224, SSL+FT, mmseg)	61.6	No	Masked Autoencoders Are Scalable Vision Learners	2021-11-11	Code
5	MAE (ViT-B/16, 224x224, SSL+FT)	61	No	Masked Autoencoders Are Scalable Vision Learners	2021-11-11	Code
6	SERE (ViT-S/16, 100ep, 224x224, SSL+FT, mmseg)	59.4	No	SERE: Exploring Feature Self-relation for Self-s...	2022-06-10	Code
7	SERE (ViT-S/16, 100ep, 224x224, SSL+FT)	58.9	No	SERE: Exploring Feature Self-relation for Self-s...	2022-06-10	Code
8	RF-ConvNext-Tiny (rfmerge, P4, 224x224, SUP)	51.3	No	RF-Next: Efficient Receptive Field Search for Co...	2022-06-14	Code
9	RF-ConvNext-Tiny (rfmultiple, P4, 224x224, SUP)	50.8	No	RF-Next: Efficient Receptive Field Search for Co...	2022-06-14	Code
10	RF-ConvNext-Tiny (rfsingle, P4, 224x224, SUP)	50.7	No	RF-Next: Efficient Receptive Field Search for Co...	2022-06-14	Code
11	ConvNext-Tiny (P4, 224x224, SUP)	48.7	No	A ConvNet for the 2020s	2022-01-10	Code
12	SERE (ViT-B/16, 100ep, 224x224, SSL)	48.6	No	SERE: Exploring Feature Self-relation for Self-s...	2022-06-10	Code
13	TEC (ViT-B/16, 224x224, SSL, mmseg)	46.1	No	Towards Sustainable Self-supervised Learning	2022-10-20	Code
14	TEC (ViT-B/16, 224x224, SSL)	42.9	No	Towards Sustainable Self-supervised Learning	2022-10-20	Code
15	SERE (ViT-S/16, 100ep, 224x224, SSL, mmseg)	41	No	SERE: Exploring Feature Self-relation for Self-s...	2022-06-10	Code
16	SERE (ViT-S/16, 100ep, 224x224, SSL)	41	No	SERE: Exploring Feature Self-relation for Self-s...	2022-06-10	Code
17	MAE (ViT-B/16, 224x224, SSL, mmseg)	40	No	Masked Autoencoders Are Scalable Vision Learners	2021-11-11	Code
18	MAE (ViT-B/16, 224x224, SSL)	38.3	No	Masked Autoencoders Are Scalable Vision Learners	2021-11-11	Code
19	PASS (ResNet-50 D16, 224x224, LUSS)	21.6	No	Large-scale Unsupervised Semantic Segmentation	2021-06-06	Code
20	PASS (ResNet-50 D32, 224x224, LUSS)	21	No	Large-scale Unsupervised Semantic Segmentation	2021-06-06	Code
21	PASS	11.5	No	Large-scale Unsupervised Semantic Segmentation	2021-06-06	Code

#1TEC (ViT-B/16, 224x224, SSL+FT, mmseg)SOTA
63.2
mIoU (val)· 2022-10-20
Towards Sustainable Self-supervised Learning Code
#2SERE (ViT-B/16, 100ep, 224x224, SSL+FT)SOTA
63
mIoU (val)· 2022-06-10
SERE: Exploring Feature Self-relation for Self-supervised Transformer Code
#3TEC (ViT-B/16, 224x224, SSL+FT)
62
mIoU (val)· 2022-10-20
Towards Sustainable Self-supervised Learning Code
#4MAE (ViT-B/16, 224x224, SSL+FT, mmseg)SOTA
61.6
mIoU (val)· 2021-11-11
Masked Autoencoders Are Scalable Vision Learners Code
#5MAE (ViT-B/16, 224x224, SSL+FT)
61
mIoU (val)· 2021-11-11
Masked Autoencoders Are Scalable Vision Learners Code
#6SERE (ViT-S/16, 100ep, 224x224, SSL+FT, mmseg)
59.4
mIoU (val)· 2022-06-10
SERE: Exploring Feature Self-relation for Self-supervised Transformer Code
#7SERE (ViT-S/16, 100ep, 224x224, SSL+FT)
58.9
mIoU (val)· 2022-06-10
SERE: Exploring Feature Self-relation for Self-supervised Transformer Code
#8RF-ConvNext-Tiny (rfmerge, P4, 224x224, SUP)
51.3
mIoU (val)· 2022-06-14
RF-Next: Efficient Receptive Field Search for Convolutional Neural Networks Code
#9RF-ConvNext-Tiny (rfmultiple, P4, 224x224, SUP)
50.8
mIoU (val)· 2022-06-14
RF-Next: Efficient Receptive Field Search for Convolutional Neural Networks Code
#10RF-ConvNext-Tiny (rfsingle, P4, 224x224, SUP)
50.7
mIoU (val)· 2022-06-14
RF-Next: Efficient Receptive Field Search for Convolutional Neural Networks Code
#11ConvNext-Tiny (P4, 224x224, SUP)
48.7
mIoU (val)· 2022-01-10
A ConvNet for the 2020s Code
#12SERE (ViT-B/16, 100ep, 224x224, SSL)
48.6
mIoU (val)· 2022-06-10
SERE: Exploring Feature Self-relation for Self-supervised Transformer Code
#13TEC (ViT-B/16, 224x224, SSL, mmseg)
46.1
mIoU (val)· 2022-10-20
Towards Sustainable Self-supervised Learning Code
#14TEC (ViT-B/16, 224x224, SSL)
42.9
mIoU (val)· 2022-10-20
Towards Sustainable Self-supervised Learning Code
#15SERE (ViT-S/16, 100ep, 224x224, SSL, mmseg)
41
mIoU (val)· 2022-06-10
SERE: Exploring Feature Self-relation for Self-supervised Transformer Code
#16SERE (ViT-S/16, 100ep, 224x224, SSL)
41
mIoU (val)· 2022-06-10
SERE: Exploring Feature Self-relation for Self-supervised Transformer Code
#17MAE (ViT-B/16, 224x224, SSL, mmseg)
40
mIoU (val)· 2021-11-11
Masked Autoencoders Are Scalable Vision Learners Code
#18MAE (ViT-B/16, 224x224, SSL)
38.3
mIoU (val)· 2021-11-11
Masked Autoencoders Are Scalable Vision Learners Code
#19PASS (ResNet-50 D16, 224x224, LUSS)SOTA
21.6
mIoU (val)· 2021-06-06
Large-scale Unsupervised Semantic Segmentation Code
#20PASS (ResNet-50 D32, 224x224, LUSS)
21
mIoU (val)· 2021-06-06
Large-scale Unsupervised Semantic Segmentation Code
#21PASS
11.5
mIoU (val)· 2021-06-06
Large-scale Unsupervised Semantic Segmentation Code