Research map / Video / Video segmentation and tracking
Recurrent Dynamic Embedding for Video Object Segmentation
Check 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 constant-size RDE memory instead of an ever-growing memory bank | RDE-VOS · DAVIS 2017 validation | J&F: 85.3 -> 86.1 | Helped |
| Use constant-size RDE memory instead of an ever-growing memory bank | RDE-VOS · DAVIS 2017 validation | FPS: 20.2 -> 27.0 | Helped |
| Use constant-size RDE memory instead of an ever-growing memory bank | RDE-VOS · DAVIS 2016 validation | J&F: 91.7 -> 91.6 | No clear effect |
| Use RDE-VOS with BL30K training data | RDE-VOS · YouTube-VOS 2019 validation | Overall: 84.2 -> 83.3 | Hurt |
| Add RDE to the first-frame embedding (ablation) | RDE-VOS · DAVIS 2017 validation | J&F: 71.6 -> 85.3 | Helped |
| Add RDE to the latest-frame embedding (ablation) | RDE-VOS · DAVIS 2017 validation | J&F: 80.4 -> 82.2 | Helped |
| Add RDE to first- and latest-frame embeddings (ablation) | RDE-VOS · DAVIS 2017 validation | J&F: 84.6 -> 85.4 | Helped |
| Sample the GT-frame embedding twice alongside latest-frame embedding and RDE (ablation) | RDE-VOS · DAVIS 2017 validation | J&F: 85.4 -> 86.1 | Helped |
| Set RDE update sampling interval θ to 3 rather than 2, 4, or 5 (ablation) | RDE-VOS · DAVIS 2017 validation | J&F: 85.1; 86.1; 85.1; 84.2 | Mixed |
| Add mask consistency loss L_MC (ablation) | RDE-VOS · DAVIS 2017 validation | J&F: 83.7 -> 84.2 | Helped |
| Add unbiased guidance loss L_UG (ablation) | RDE-VOS · DAVIS 2017 validation | J&F: 82.9 -> 84.2 | Helped |
| Add both mask consistency and unbiased guidance losses (ablation) | RDE-VOS · DAVIS 2017 validation | J&F: 82.5 -> 84.2 | Helped |
Closest papers by meaning
- Efficient Video Object Segmentation and Tracking with Recurrent Dynamic Submodel (CVPR 2026 · similarity 0.90)
- RVOS: End-To-End Recurrent Network for Video Object Segmentation (CVPR 2019 · similarity 0.90)
- LiVOS: Light Video Object Segmentation with Gated Linear Matching (CVPR 2025 · similarity 0.89)
- RMem: Restricted Memory Banks Improve Video Object Segmentation (CVPR 2024 · similarity 0.88)
- SAM-I2V: Upgrading SAM to Support Promptable Video Segmentation with Less than 0.2% Training Cost (CVPR 2025 · similarity 0.87)
- Learning Spatial-Semantic Features for Robust Video Object Segmentation (ICLR 2025 · similarity 0.87)
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.