Video Object Segmentation Papers

[nijkah]

Applications

• Semi-supervised Video Object Segmentation
  • The target information is given only as a mask in the first frame.
• Interactive Video Object Segmentation
  • It receives two inputs, indicating positive and negative samples) as scribbles.
• Unsupervised Video Object Segmentation
  • This task should segment all salient objects
• Video Instance Segmentation
  • simultaneous detection, segmentation and tracking of object instances in videos
  • should predict the class of objects
• Video Object Segmentation with Bounding Box(Weakly&Semi-supervised Video Object Segmentation)
  • The target information is given only as a bounding box in the first frame.

Paper

One-Shot Video Object Segmentation

• paper
• code

Learning Video Object Segmentation from Static Images

• paper
• code baseline code - MaskTrack_Box

Fast Online Object Tracking and Segmentation: A Unifying Approach

• paper
• code

BoLTVOS: Box-Level Tracking for Video Object Segmentation

• paper

References

DATASETS

DAVIS Dataset

• ref DAVIS-2016(Single Object), DAVIS-2017(Multiple Object)

Youtube-VOS

• ref