Video Instance Segmentation on YouTube-VIS 2021

Metric: AP50 (higher is better)

LeaderboardDataset

Loading chart...

Results

Hide extra data

Sort:

#	Model↕	AP50▼	Extra Data	Paper	Date↕	Code
1	CAVIS(VIT-L, Offline)	87.3	Yes	Context-Aware Video Instance Segmentation	2024-07-03	Code
2	DVIS++(VIT-L, Offline)	86.7	Yes	DVIS++: Improved Decoupled Framework for Univers...	2023-12-20	Code
3	DVIS-DAQ(VIT-L, Offline)	86.1	Yes	DVIS-DAQ: Improving Video Segmentation via Dynam...	2024-03-29	Code
4	RefineVIS (Swin-L, online)	84.1	Yes	RefineVIS: Video Instance Segmentation with Temp...	2023-06-07	-
5	DVIS(Swin-L)	83	Yes	DVIS: Decoupled Video Instance Segmentation Fram...	2023-06-06	Code
6	DVIS++(VIT-L, Online)	82.7	Yes	DVIS++: Improved Decoupled Framework for Univers...	2023-12-20	Code
7	NOVIS (Swin-L)	82	Yes	NOVIS: A Case for End-to-End Near-Online Video I...	2023-08-29	-
8	TarViS (Swin-L)	81.4	Yes	TarViS: A Unified Approach for Target-based Vide...	2023-01-06	Code
9	GRAtt-VIS (Swin-L)	81.3	Yes	GRAtt-VIS: Gated Residual Attention for Auto Rec...	2023-05-26	Code
10	GenVIS (Swin-L)	80.9	Yes	A Generalized Framework for Video Instance Segme...	2022-11-16	Code
11	IDOL (Swin-L)	80.8	No	In Defense of Online Models for Video Instance S...	2022-07-21	Code
12	MDQE(Swin-L)	80.7	No	MDQE: Mining Discriminative Query Embeddings to ...	2023-03-25	Code
13	VITA (Swin-L)	80.6	Yes	VITA: Video Instance Segmentation via Object Tok...	2022-06-09	Code
14	Tube-Link(Swin-L)	79.4	No	Tube-Link: A Flexible Cross Tube Framework for U...	2023-03-22	Code
15	UniVS(Swin-L)	79.4	Yes	UniVS: Unified and Universal Video Segmentation ...	2024-02-28	Code
16	DeVIS (Swin-L)	77.7	No	DeVIS: Making Deformable Transformers Work for V...	2022-07-22	Code
17	MinVIS (Swin-L)	76.6	No	MinVIS: A Minimal Video Instance Segmentation Fr...	2022-08-03	Code
18	BoxVIS(Swin-L & Box-sup)	76.4	No	BoxVIS: Video Instance Segmentation with Box Ann...	2023-03-26	Code
19	InstanceFormer (Swin-L)	73.7	Yes	InstanceFormer: An Online Video Instance Segment...	2022-08-22	Code
20	TarViS (Swin-T)	71.6	Yes	TarViS: A Unified Approach for Target-based Vide...	2023-01-06	Code
21	TarViS (ResNet-50)	69.6	Yes	TarViS: A Unified Approach for Target-based Vide...	2023-01-06	Code
22	NOVIS (ResNet-50)	69.4	Yes	NOVIS: A Case for End-to-End Near-Online Video I...	2023-08-29	-
23	GRAtt-VIS (ResNet-50)	69.2	Yes	GRAtt-VIS: Gated Residual Attention for Auto Rec...	2023-05-26	Code
24	DeVIS (ResNet-50)	66.8	No	DeVIS: Making Deformable Transformers Work for V...	2022-07-22	Code
25	InstanceFormer (ResNet-50)	62.4	Yes	InstanceFormer: An Online Video Instance Segment...	2022-08-22	Code
26	STMask(R101-DCN-FPN)	54	No	Spatial Feature Calibration and Temporal Fusion ...	2021-04-06	Code

#1CAVIS(VIT-L, Offline)SOTA
87.3
AP50· Extra Data· 2024-07-03
Context-Aware Video Instance Segmentation Code
#2DVIS++(VIT-L, Offline)SOTA
86.7
AP50· Extra Data· 2023-12-20
DVIS++: Improved Decoupled Framework for Universal Video Segmentation Code
#3DVIS-DAQ(VIT-L, Offline)
86.1
AP50· Extra Data· 2024-03-29
DVIS-DAQ: Improving Video Segmentation via Dynamic Anchor Queries Code
#4RefineVIS (Swin-L, online)SOTA
84.1
AP50· Extra Data· 2023-06-07
RefineVIS: Video Instance Segmentation with Temporal Attention Refinement
#5DVIS(Swin-L)SOTA
83
AP50· Extra Data· 2023-06-06
DVIS: Decoupled Video Instance Segmentation Framework Code
#6DVIS++(VIT-L, Online)
82.7
AP50· Extra Data· 2023-12-20
DVIS++: Improved Decoupled Framework for Universal Video Segmentation Code
#7NOVIS (Swin-L)
82
AP50· Extra Data· 2023-08-29
NOVIS: A Case for End-to-End Near-Online Video Instance Segmentation
#8TarViS (Swin-L)SOTA
81.4
AP50· Extra Data· 2023-01-06
TarViS: A Unified Approach for Target-based Video Segmentation Code
#9GRAtt-VIS (Swin-L)
81.3
AP50· Extra Data· 2023-05-26
GRAtt-VIS: Gated Residual Attention for Auto Rectifying Video Instance Segmentation Code
#10GenVIS (Swin-L)SOTA
80.9
AP50· Extra Data· 2022-11-16
A Generalized Framework for Video Instance Segmentation Code
#11IDOL (Swin-L)SOTA
80.8
AP50· 2022-07-21
In Defense of Online Models for Video Instance Segmentation Code
#12MDQE(Swin-L)
80.7
AP50· 2023-03-25
MDQE: Mining Discriminative Query Embeddings to Segment Occluded Instances on Challenging Videos Code
#13VITA (Swin-L)SOTA
80.6
AP50· Extra Data· 2022-06-09
VITA: Video Instance Segmentation via Object Token Association Code
#14Tube-Link(Swin-L)
79.4
AP50· 2023-03-22
Tube-Link: A Flexible Cross Tube Framework for Universal Video Segmentation Code
#15UniVS(Swin-L)
79.4
AP50· Extra Data· 2024-02-28
UniVS: Unified and Universal Video Segmentation with Prompts as Queries Code
#16DeVIS (Swin-L)
77.7
AP50· 2022-07-22
DeVIS: Making Deformable Transformers Work for Video Instance Segmentation Code
#17MinVIS (Swin-L)
76.6
AP50· 2022-08-03
MinVIS: A Minimal Video Instance Segmentation Framework without Video-based Training Code
#18BoxVIS(Swin-L & Box-sup)
76.4
AP50· 2023-03-26
BoxVIS: Video Instance Segmentation with Box Annotations Code
#19InstanceFormer (Swin-L)
73.7
AP50· Extra Data· 2022-08-22
InstanceFormer: An Online Video Instance Segmentation Framework Code
#20TarViS (Swin-T)
71.6
AP50· Extra Data· 2023-01-06
TarViS: A Unified Approach for Target-based Video Segmentation Code
#21TarViS (ResNet-50)
69.6
AP50· Extra Data· 2023-01-06
TarViS: A Unified Approach for Target-based Video Segmentation Code
#22NOVIS (ResNet-50)
69.4
AP50· Extra Data· 2023-08-29
NOVIS: A Case for End-to-End Near-Online Video Instance Segmentation
#23GRAtt-VIS (ResNet-50)
69.2
AP50· Extra Data· 2023-05-26
GRAtt-VIS: Gated Residual Attention for Auto Rectifying Video Instance Segmentation Code
#24DeVIS (ResNet-50)
66.8
AP50· 2022-07-22
DeVIS: Making Deformable Transformers Work for Video Instance Segmentation Code
#25InstanceFormer (ResNet-50)
62.4
AP50· Extra Data· 2022-08-22
InstanceFormer: An Online Video Instance Segmentation Framework Code
#26STMask(R101-DCN-FPN)SOTA
54
AP50· 2021-04-06
Spatial Feature Calibration and Temporal Fusion for Effective One-stage Video Instance Segmentation Code