Instance Segmentation on ADE20K val

Metric: AP (higher is better)

LeaderboardDataset

Loading chart...

Results

Hide extra data

Sort:

#	Model↕	AP▼	Extra Data	Paper	Date↕	Code
1	OneFormer (InternImage-H, emb_dim=1024, single-scale, 896x896, COCO-Pretrained)	44.2	No	OneFormer: One Transformer to Rule Universal Ima...	2022-11-10	Code
2	OpenSeeD	42.6	Yes	A Simple Framework for Open-Vocabulary Segmentat...	2023-03-14	Code
3	ViT-P (OneFormer, DiNAT-L, single-scale, 1280x1280, COCO_pretrain)	40.7	Yes	The Missing Point in Vision Transformers for Uni...	2025-05-26	Code
4	OneFormer (DiNAT-L, single-scale, 1280x1280, COCO-pretrain)	40.2	Yes	OneFormer: One Transformer to Rule Universal Ima...	2022-11-10	Code
5	X-Decoder (Davit-d5, Deform, single-scale, 1280x1280)	38.7	Yes	Generalized Decoding for Pixel, Image, and Langu...	2022-12-21	Code
6	ViT-P (OneFormer, DiNAT-L, single-scale, 1280x1280)	37.8	No	The Missing Point in Vision Transformers for Uni...	2025-05-26	Code
7	OneFormer (DiNAT-L, single-scale)	36	No	OneFormer: One Transformer to Rule Universal Ima...	2022-11-10	Code
8	OneFormer (Swin-L, single-scale)	35.9	No	OneFormer: One Transformer to Rule Universal Ima...	2022-11-10	Code
9	X-Decoder (L)	35.8	Yes	Generalized Decoding for Pixel, Image, and Langu...	2022-12-21	Code
10	DiNAT-L (Mask2Former, single-scale)	35.4	No	Dilated Neighborhood Attention Transformer	2022-09-29	Code
11	Mask2Former (Swin-L, single-scale)	34.9	No	Masked-attention Mask Transformer for Universal ...	2021-12-02	Code
12	Mask2Former (Swin-L + FAPN)	33.4	No	Masked-attention Mask Transformer for Universal ...	2021-12-02	Code
13	Mask2Former (ResNet50)	26.4	No	Masked-attention Mask Transformer for Universal ...	2021-12-02	Code

#1OneFormer (InternImage-H, emb_dim=1024, single-scale, 896x896, COCO-Pretrained)SOTA
44.2
AP· 2022-11-10
OneFormer: One Transformer to Rule Universal Image Segmentation Code
#2OpenSeeD
42.6
AP· Extra Data· 2023-03-14
A Simple Framework for Open-Vocabulary Segmentation and Detection Code
#3ViT-P (OneFormer, DiNAT-L, single-scale, 1280x1280, COCO_pretrain)
40.7
AP· Extra Data· 2025-05-26
The Missing Point in Vision Transformers for Universal Image Segmentation Code
#4OneFormer (DiNAT-L, single-scale, 1280x1280, COCO-pretrain)
40.2
AP· Extra Data· 2022-11-10
OneFormer: One Transformer to Rule Universal Image Segmentation Code
#5X-Decoder (Davit-d5, Deform, single-scale, 1280x1280)
38.7
AP· Extra Data· 2022-12-21
Generalized Decoding for Pixel, Image, and Language Code
#6ViT-P (OneFormer, DiNAT-L, single-scale, 1280x1280)
37.8
AP· 2025-05-26
The Missing Point in Vision Transformers for Universal Image Segmentation Code
#7OneFormer (DiNAT-L, single-scale)
36
AP· 2022-11-10
OneFormer: One Transformer to Rule Universal Image Segmentation Code
#8OneFormer (Swin-L, single-scale)
35.9
AP· 2022-11-10
OneFormer: One Transformer to Rule Universal Image Segmentation Code
#9X-Decoder (L)
35.8
AP· Extra Data· 2022-12-21
Generalized Decoding for Pixel, Image, and Language Code
#10DiNAT-L (Mask2Former, single-scale)SOTA
35.4
AP· 2022-09-29
Dilated Neighborhood Attention Transformer Code
#11Mask2Former (Swin-L, single-scale)SOTA
34.9
AP· 2021-12-02
Masked-attention Mask Transformer for Universal Image Segmentation Code
#12Mask2Former (Swin-L + FAPN)
33.4
AP· 2021-12-02
Masked-attention Mask Transformer for Universal Image Segmentation Code
#13Mask2Former (ResNet50)
26.4
AP· 2021-12-02
Masked-attention Mask Transformer for Universal Image Segmentation Code