Research map / Video / Video segmentation and tracking
Iteratively Selecting an Easy Reference Frame Makes Unsupervised Video Object Segmentation Easier
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Select easy reference frames and iteratively update their saliency cues with temporal information | UVOS using STM · DAVIS16 validation | J Mean: 82.5 -> 84.5 | Helped |
| Select easy reference frames and iteratively update their saliency cues with temporal information | UVOS using STM · DAVIS16 validation | F Mean: 81.2 -> 86.7 | Helped |
| Use the IMP framework (ablation) | UVOS using COSNet initial masks · DAVIS16 | J&F: +2.6 | Helped |
| Use the IMP framework (ablation) | UVOS using MATNet initial masks · DAVIS16 | J&F: +5.2 | Helped |
| Use 2 easy frames instead of 1 (ablation) | IMP · DAVIS16 | J&F: 84.7 -> 85.5 | Helped |
| Use 3 rather than 2 easy frames (ablation) | IMP · DAVIS16 | J&F: 85.5 -> 85.1 | Hurt |
| Use 4 rather than 2 easy frames (ablation) | IMP · DAVIS16 | J&F: 85.5 -> 85.0 | Hurt |
| Estimate mask quality with structural similarity S instead of MAE (ablation) | Easy Frame Selector · DAVIS16 | J&F: 80.7 -> 85.5 | Helped |
| Add Temporal Information Updating (ablation) | IMP with 4 iterations · DAVIS16 | J&F: 83.5 -> 85.5 | Helped |
| Add Temporal Information Updating (ablation) | IMP with 1 iteration · DAVIS16 | J&F: 78.5 -> 78.5 | No clear effect |
| Replace STM with RGMP as the mask propagation model (ablation) | IMP · DAVIS16, 4 iterations | J&F: 85.5 -> 59.2 | Hurt |
| Replace STM with KMN as the mask propagation model (ablation) | IMP · DAVIS16, 4 iterations | J&F: 85.5 -> 84.1 | Hurt |
Closest papers by meaning
- Make One-Shot Video Object Segmentation Efficient Again (NeurIPS 2020 · similarity 0.89)
- Mask-Free Video Instance Segmentation (CVPR 2023 · similarity 0.88)
- Reciprocal Transformations for Unsupervised Video Object Segmentation (CVPR 2021 · similarity 0.88)
- Segment Anything Across Shots: A Method and Benchmark (AAAI 2026 · similarity 0.88)
- LiVOS: Light Video Object Segmentation with Gated Linear Matching (CVPR 2025 · similarity 0.87)
- Audio-Visual Segmentation via Unlabeled Frame Exploitation (CVPR 2024 · similarity 0.86)
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.