AI Paper Atlas

Research map / Video / Video segmentation and tracking

SWEM: Towards Real-Time Video Object Segmentation With Sequential Weighted Expectation-Maximization

Zhihui Lin, Tianyu Yang, Maomao Li, Ziyu Wang, Chun Yuan, Wenhao Jiang, Wei Liu

CVPR 2022 · Video › Video segmentation and tracking · approach: Mask propagation for unsupervised video segmentation · 44 citations (OpenAlex)

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use adaptive pixel weights rather than fixed mask weights (ablation)SWEM · DAVIS 2017 valJ&F: 77.6 -> 81.9Improved
Use 128 rather than 32 bases (ablation)SWEM · DAVIS 2017 valJ&F: 80.2 -> 81.9Improved
Increase the number of bases from 128 to 256 (ablation)SWEM · DAVIS 2017 valJ&F: 81.9 -> 82.0No clear effect
Increase SWEM iterations from 1 to 4 (ablation)SWEM · DAVIS 2017 valJ&F: 77.9 -> 81.9Improved
Increase SWEM iterations from 4 to 7 (ablation)SWEM · DAVIS 2017 valJ&F: 81.9 -> 79.8Hurt
Use 64 rather than 8 permutation-invariant features (ablation)SWEM · DAVIS 2017 valJ&F: 81.0 -> 81.9Improved
Increase permutation-invariant features from 64 to 128 (ablation)SWEM · DAVIS 2017 valJ&F: 81.9 -> 81.6Hurt
Set the similarity temperature to τ=0.05 rather than τ=1 (ablation)SWEM · DAVIS 2017 valJ&F: 66.7 -> 81.9Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy