Unsupervised Video Object Segmentation On 10

评估指标

评测结果

各个模型在此基准测试上的表现结果

				Paper Title	Repository
DEVA (DIS)	90.2	88.9	87.6	Tracking Anything with Decoupled Video Segmentation
GSANet	89.6	88.9	88.3	Guided Slot Attention for Unsupervised Video Object Segmentation
FakeFlow	89.0	88.5	88.0	Improving Unsupervised Video Object Segmentation via Fake Flow Generation	-
DPA	88.4	87.6	86.8	Dual Prototype Attention for Unsupervised Video Object Segmentation
TMO (MiT-b1)	87.8	87.2	86.6	Treating Motion as Option to Reduce Motion Dependency in Unsupervised Video Object Segmentation
AMP	87.5	87.3	87.1	Adaptive Multi-source Predictor for Zero-shot Video Object Segmentation
IMP	86.7	85.6	84.5	Iteratively Selecting an Easy Reference Frame Makes Unsupervised Video Object Segmentation Easier	-
DLDA	86.6	85.75	84.9	Self-supervised Video Object Segmentation with Distillation Learning of Deformable Attention	-
TMO (RN-101)	86.6	86.1	85.6	Treating Motion as Option to Reduce Motion Dependency in Unsupervised Video Object Segmentation
D2Conv3D	86.5	86.0	85.5	D2Conv3D: Dynamic Dilated Convolutions for Object Segmentation in Videos	-
PMN	86.4	85.9	85.4	Unsupervised Video Object Segmentation via Prototype Memory Network
TransportNet	85.0	84.8	84.5	Deep Transport Network for Unsupervised Video Object Segmentation	-
RTNet	84.7	85.2	85.6	Reciprocal Transformations for Unsupervised Video Object Segmentation	-
3DC-Seg	84.7	84.5	84.3	Making a Case for 3D Convolutions for Object Segmentation in Videos
AMC-Net	84.6	84.6	84.5	Learning Motion-Appearance Co-Attention for Zero-Shot Video Object Segmentation	-
FSNet	83.1	83.3	83.4	Full-Duplex Strategy for Video Object Segmentation
DFNet	81.8	82.6	83.4	Learning Discriminative Feature with CRF for Unsupervised Video Object Segmentation	-
MATNet	80.7	81.6	82.4	Motion-Attentive Transition for Zero-Shot Video Object Segmentation
WCS-Net	80.7	81.5	82.2	Unsupervised Video Object Segmentation with Joint Hotspot Tracking	-
AD-Net	80.5	81.1	81.7	Anchor Diffusion for Unsupervised Video Object Segmentation

0 of 25 row(s) selected.