Video Instance Segmentation On Youtube Vis 1

评估指标

AP50

AP75

AR1

AR10

mask AP

评测结果

各个模型在此基准测试上的表现结果

						Paper Title	Repository
DVIS++(VIT-L, Online)	88.8	75.3	57.9	73.7	67.7	DVIS++: Improved Decoupled Framework for Universal Video Segmentation
DVIS	88.0	72.7	56.5	70.3	64.9	DVIS: Decoupled Video Instance Segmentation Framework
Tube-Link	86.6	71.3	55.9	69.1	64.6	Tube-Link: A Flexible Cross Tube Framework for Universal Video Segmentation
MinVIS (Swin-L)	83.3	68.6	54.8	66.6	61.6	MinVIS: A Minimal Video Instance Segmentation Framework without Video-based Training
Mask2Former (Swin-L)	84.4	67.0	-	-	60.4	Mask2Former for Video Instance Segmentation
UniVS(Swin-L)	82.1	65.3	54.7	66.8	60.0	UniVS: Unified and Universal Video Segmentation with Prompts as Queries
MDQE(Swin-L)	84.9	67.3	53.5	65.0	59.9	MDQE: Mining Discriminative Query Embeddings to Segment Occluded Instances on Challenging Videos
SeqFormer (Swin-L)	82.1	66.4	51.7	64.4	59.3	SeqFormer: Sequential Transformer for Video Instance Segmentation
DeVIS (Swin-L)	80.8	66.3	50.8	61.0	57.1	DeVIS: Making Deformable Transformers Work for Video Instance Segmentation
InstanceFormer(Swin-L)	78.0	64.2	50.9	61.6	56.3	InstanceFormer: An Online Video Instance Segmentation Framework
TCIS (Swin-S)	76.6	65.6	47	57.9	54.3	1st Place Solution for YouTubeVOS Challenge 2021:Video Instance Segmentation	-
Video K-Net (Swin-Base)	79.0	59.6	49.7	59.9	54.1	Video K-Net: A Simple, Strong, and Unified Baseline for Video Segmentation
NOVIS (ResNet-50)	75.7	56.9	50.3	60.6	52.8	NOVIS: A Case for End-to-End Near-Online Video Instance Segmentation	-
IDOL (ResNet-50)	74	52.9	47.7	58.7	49.5	In Defense of Online Models for Video Instance Segmentation
Mask2Former (ResNet-101)	72.8	54.2	-	-	49.2	Mask2Former for Video Instance Segmentation
SeqFormer (ResNet-101)	71.1	55.7	46.8	56.9	49.0	SeqFormer: Sequential Transformer for Video Instance Segmentation
MSN	69.4	54.9	40.1	55.0	48.8	MSN: Efficient Online Mask Selection Network for Video Instance Segmentation
SeqFormer (ResNet-50)	69.8	51.8	45.5	54.8	47.4	SeqFormer: Sequential Transformer for Video Instance Segmentation
Mask2Former (ResNet-50)	68.0	50.0	-	-	46.4	Mask2Former for Video Instance Segmentation
InstanceFormer(ResNet-50)	68.6	49.6	42.1	53.5	45.6	InstanceFormer: An Online Video Instance Segmentation Framework

0 of 43 row(s) selected.

Command Palette

Video Instance Segmentation On Youtube Vis 1

评估指标

评测结果