Research map / Video / Video segmentation and tracking
Collaborative Video Object Segmentation by Multi-Scale Foreground-Background Integration
Zongxin Yang, Yunchao Wei, Yi Yang
Read the paperCheck your idea against 137,217 papers
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add background embedding (foreground and background) to matching and attention (ablation) | CFBI · DAVIS-2017 val | Avg J&F: 70.9 -> 74.9 | Improved |
| add background embedding to pixel-level matching (ablation) | CFBI · DAVIS-2017 val | Avg J&F: 73.0 -> 74.9 | Improved |
| add background embedding to instance-level attention (ablation) | CFBI · DAVIS-2017 val | Avg J&F: 72.3 -> 74.9 | Improved |
| add foreground and background bias b_F, b_B to distance metric (ablation) | CFBI · DAVIS-2017 val | Avg J&F: 72.8 -> 74.9 | Improved |
| use multi-local windows instead of one local window (ablation) | CFBI · DAVIS-2017 val | Avg J&F: 73.8 -> 74.9 | Improved |
| add sequential training with predicted previous masks (ablation) | CFBI · DAVIS-2017 val | Avg J&F: 73.3 -> 74.9 | Improved |
| use collaborative ensembler instead of 4 separable conv layers (ablation) | CFBI · DAVIS-2017 val | Avg J&F: 73.3 -> 74.9 | Improved |
| add balanced random-crop (ablation) | CFBI · DAVIS-2017 val | Avg J&F: 72.8 -> 74.9 | Improved |
| add instance-level attention guidance to the ensembler (ablation) | CFBI · DAVIS-2017 val | Avg J&F: 72.7 -> 74.9 | Improved |
| atrous factor 2 in global matching (ablation) | CFBI · YouTube-VOS 2018 val | Avg J&F: 81.4 -> 81.3 | No clear effect |
| increase atrous factor from 1 to 4 in multi-local matching (ablation) | CFBI · YouTube-VOS 2018 val | Avg J&F: 81.4 -> 79.5 | Hurt |
| do matching at stride 4 instead of stride 16 (ablation) | CFBI · YouTube-VOS 2018 val | Avg J&F: 78.3 -> 81.6 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- FEELVOS — FEELVOS: Fast End-To-End Embedding Learning for Video Object Segmentation (CVPR 2019) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Collaborative Video Object Segmentation by Foreground-Background Integration (ECCV 2020 · similarity 0.98)
- F2Net: Learning to Focus on the Foreground for Unsupervised Video Object Segmentation (AAAI 2021 · similarity 0.87)
- Video Object Segmentation Using Global and Instance Embedding Learning (CVPR 2021 · similarity 0.87)
- Boosting Video Object Segmentation via Space-Time Correspondence Learning (CVPR 2023 · similarity 0.87)
- JointFormer: A Unified Framework With Joint Modeling for Video Object Segmentation (TPAMI 2025 · similarity 0.87)
- FOCUS: Towards Universal Foreground Segmentation (AAAI 2025 · similarity 0.85)
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.
