Research map / Video / Video segmentation and tracking
Motion-Guided Cascaded Refinement Network for 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 |
|---|---|---|---|
| add active-contour motion guidance and CRN refinement instead of using CRN alone (ablation) | Motion-Guided CRN · DAVIS2016 validation | mIoU: 0.824 -> 0.844 | Helped |
| add CRN refinement to active-contour segmentation (ablation) | Motion-Guided CRN · DAVIS2016 validation | mIoU: 0.553 -> 0.844 | Helped |
| add offline training on DAVIS2016 after PASCAL VOC training (ablation) | CRN · DAVIS2016 validation | J Mean: 0.526 -> 0.764 | Helped |
| add online training on the first frame after offline training (ablation) | CRN · DAVIS2016 validation | J Mean: 0.764 -> 0.844 | Helped |
| replace normal residual units with SRAM attention modules and use the guidance map (ablation) | CRN · DAVIS2016 validation | J Mean: 0.805 -> 0.844 | Helped |
| increase active-contour iterations from 10 to 20 (ablation) | Active contour + CRN · DAVIS2016 validation | mIoU: 0.844 -> 0.844 | No clear effect |
| increase online-training iterations from 100 to 500 (ablation) | CRN · DAVIS2016 validation | mIoU: 0.844 -> 0.843 | Hurt |
| use 500 online-training iterations instead of 10 (ablation) | CRN · DAVIS2016 validation | time(s/f): 0.40 -> 2.18 | Hurt |
| apply Motion-Guided CRN rather than OnAVOS with CRF | DAVIS2016 validation, semi-supervised · DAVIS2016 validation | J Mean: 0.861 -> 0.844 | Hurt |
| apply Motion-Guided CRN rather than OnAVOS with CRF | DAVIS2016 validation, semi-supervised · DAVIS2016 validation | F Mean: 0.849 -> 0.857 | Helped |
| apply Motion-Guided CRN rather than OnAVOS with CRF | DAVIS2016 validation, semi-supervised · DAVIS2016 validation | time(s/f): 15.57 -> 0.73 | Helped |
| apply Motion-Guided CRN rather than OnAVOS | YoutubeObjects · YoutubeObjects | mIoU: 0.774 -> 0.766 | Hurt |
Closest papers by meaning
- Efficient Video Object Segmentation via Network Modulation (CVPR 2018 · similarity 0.88)
- Learning Video Object Segmentation From Static Images (CVPR 2017 · similarity 0.88)
- Dynamic Video Segmentation Network (CVPR 2018 · similarity 0.87)
- Learning Dynamic Network Using a Reuse Gate Function in Semi-Supervised Video Object Segmentation (CVPR 2021 · similarity 0.87)
- LiVOS: Light Video Object Segmentation with Gated Linear Matching (CVPR 2025 · similarity 0.86)
- Putting the Object Back into Video Object Segmentation (CVPR 2024 · 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.