Research map / Video / Video segmentation and tracking
Video Object Segmentation Using Space-Time Memory Networks
Seoung Wug Oh, Joon-Young Lee, Ning Xu, Seon Joo Kim
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 pre-training on simulated videos from static images (ablation) | STM main training · Youtube-VOS val | overall score: 68.2 -> 79.4 | Improved |
| add main training on real videos after pre-training (ablation) | STM · Youtube-VOS val | overall score: 69.1 -> 79.4 | Improved |
| add pre-training on static images (DAVIS-2017 training) (ablation) | STM · DAVIS-2017 val | J: 38.1 -> 69.2 | Improved |
| add previous frame to first-frame memory (ablation) | STM · Youtube-VOS val | overall score: 68.9 -> 78.4 | Improved |
| add first frame to previous-frame memory (ablation) | STM · Youtube-VOS val | overall score: 69.7 -> 78.4 | Improved |
| add intermediate memory frames every 5 frames (ablation) | STM · Youtube-VOS val | overall score: 78.4 -> 79.4 | Improved |
| add intermediate memory frames every 5 frames (ablation) | STM · DAVIS-2016 val | J mean: 87.8 -> 88.7 | Improved |
| add intermediate memory frames every 5 frames (ablation) | STM · DAVIS-2017 val | J mean: 77.7 -> 79.2 | Improved |
| add intermediate memory frames every 5 frames: runtime (ablation) | STM · DAVIS-2016 val | time (lower is better): 0.07 -> 0.16 | Hurt |
| add Youtube-VOS training data | STM · DAVIS-2016 val | J mean: 84.8 -> 88.7 | Improved |
| add Youtube-VOS training data | STM · DAVIS-2017 val | J mean: 69.2 -> 79.2 | Improved |
| train on DAVIS only, evaluate on Youtube-VOS | STM · Youtube-VOS val | overall score: 79.4 -> 56.3 | Hurt |
Closest papers by meaning
- Space-Time Memory Networks for Video Object Segmentation With User Guidance (TPAMI 2020 · similarity 0.96)
- Video Object Segmentation Using Kernelized Memory Network With Multiple Kernels (TPAMI 2022 · similarity 0.93)
- Video Object Segmentation With Dynamic Memory Networks and Adaptive Object Alignment (ICCV 2021 · similarity 0.92)
- Learning Video Object Segmentation With Visual Memory (ICCV 2017 · similarity 0.92)
- LiVOS: Light Video Object Segmentation with Gated Linear Matching (CVPR 2025 · similarity 0.90)
- Learning Spatial-Semantic Features for Robust Video Object Segmentation (ICLR 2025 · similarity 0.90)
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.
