Research map / Video / Video segmentation and tracking
Rethinking Space-Time Networks with Improved Memory Coverage for Efficient Video Object Segmentation
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 |
|---|---|---|---|
| use L2 similarity instead of dot product for affinity (ablation) | STCN (Ck=64) · DAVIS 2017 val | J&F: 83.2 -> 85.4 | Helped |
| use cosine instead of dot product similarity (ablation) | STCN (Ck=128) · DAVIS 2017 val | J&F: 84.1 -> 82.6 | Hurt |
| reduce key dim 128 -> 64 with L2 similarity (ablation) | STCN · DAVIS 2017 val | J&F: 85.0 -> 85.4 | Helped |
| reduce key dim 128 -> 64 with dot product (ablation) | STCN · DAVIS 2017 val | J&F: 84.1 -> 83.2 | Hurt |
| drop last-frame temporary memory (every 5th only) (ablation) | STCN · DAVIS 2017 val | J&F: 83.1 -> 85.4 | Helped |
| drop last-frame temporary memory (ablation) | STCN · DAVIS 2017 val | FPS: 15.4 -> 20.2 | Helped |
| drop last-frame temporary memory | STM · DAVIS 2017 val | J&F: 82.7 -> 81.0 | Hurt |
| add static-image pretraining (ablation) | STCN · DAVIS 2017 val | J&F: 82.5 -> 85.4 | Helped |
| add BL30K synthetic training data | STCN · YouTubeVOS 2018 val | G: 83.0 -> 84.3 | Helped |
| STCN shared key encoder and L2 affinity vs STM | STM · DAVIS 2017 val | J&F: 81.8 -> 85.4 | Helped |
Closest papers by meaning
- Boosting Video Object Segmentation via Space-Time Correspondence Learning (CVPR 2023 · similarity 0.90)
- Recurrent Dynamic Embedding for Video Object Segmentation (CVPR 2022 · similarity 0.89)
- LiVOS: Light Video Object Segmentation with Gated Linear Matching (CVPR 2025 · similarity 0.88)
- Learning Spatial-Semantic Features for Robust Video Object Segmentation (ICLR 2025 · similarity 0.86)
- Learning Segmentation from Point Trajectories (NeurIPS 2024 · similarity 0.86)
- Exploiting Temporal State Space Sharing for Video Semantic Segmentation (CVPR 2025 · 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.