Research map / Detection and segmentation / Open-set and few-shot detection
Delving into the Cyclic Mechanism in Semi-supervised 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 cyclic training (ablation) | STM · DAVIS17 validation | J&F (%): 69.7 -> 71.7 | Helped |
| Add gradient correction (ablation) | STM · DAVIS17 validation | J&F (%): 69.7 -> 70.3 | Helped |
| Add cyclic training and gradient correction (ablation) | STM · DAVIS17 validation | J&F (%): 69.7 -> 72.3 | Helped |
| Add gradient correction to cyclic training | STM-cycle · DAVIS17 test-dev | J&F (%): 57.8 -> 58.6 | Helped |
| Add gradient correction to cyclic training | STM-cycle · Youtube-VOS validation | G (%): 69.9 -> 70.8 | Helped |
| Add cyclic training (ablation) | STM · DAVIS17 validation, first-frame reference | J&F (%): 65.2 -> 67.6 (+2.4) | Helped |
| Add cyclic training (ablation) | STM · DAVIS17 validation, last-frame reference | J&F (%): 56.8 -> 61.2 (+4.4) | Helped |
| Add cyclic training (ablation) | STM · DAVIS17 validation, first- and last-frame references | J&F (%): 67.3 -> 69.2 (+1.9) | Helped |
| Add cyclic training (ablation) | STM · DAVIS17 validation, MEM references | J&F (%): 69.7 -> 71.7 (+2.0) | Helped |
| Apply gradient correction to baseline-predicted low-quality memory masks (ablation) | STM · DAVIS17 validation | J&F (%): 69.1 -> 70.1 | Helped |
| Apply gradient correction to bounding-box memory masks (ablation) | STM · DAVIS17 validation | J&F (%): 64.5 -> 71.8 | Helped |
| Use all preceding frames and masks rather than only the current frame and prediction in the cycle reference (ablation) | STM-cycle · DAVIS17 validation | J&F (%): 71.7 -> 71.6 | No clear effect |
Closest papers by meaning
- Learning Dynamic Network Using a Reuse Gate Function in Semi-Supervised Video Object Segmentation (CVPR 2021 · similarity 0.89)
- Fast Video Object Segmentation by Reference-Guided Mask Propagation (CVPR 2018 · similarity 0.88)
- Weakly Supervised Instance Segmentation for Videos With Temporal Mask Consistency (CVPR 2021 · similarity 0.88)
- A Transductive Approach for Video Object Segmentation (CVPR 2020 · similarity 0.88)
- Self-supervised Object-Centric Learning for Videos (NeurIPS 2023 · similarity 0.88)
- Holistic Correction with Object Prototype for Video Object Segmentation (AAAI 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.