Research map / Video / Video segmentation and tracking
Exploring the Semi-Supervised Video Object Segmentation Problem from a Cyclic Perspective
Yuxi Li, Ning Xu, Wenjie Yang, John See, Weiyao Lin
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 cycle-consistency training loss (ablation) | STM ResNet50 · DAVIS17 validation | J&F: 69.7 -> 71.7 | Improved |
| add cycle-consistency training loss (ablation) | STM ResNet50 · DAVIS16 validation | J&F: 82.3 -> 83.9 | Improved |
| add gradient correction at inference on top of cyclic model (ablation) | STM-ResNet50-cycle · DAVIS17 validation | J&F: 71.7 -> 72.9 | Improved |
| add gradient correction at inference on top of cyclic model (ablation) | STM-ResNet50-cycle · DAVIS16 validation | J&F: 83.9 -> 84.0 | No clear effect |
| add cyclic training when reference is only the last predicted frame (ablation) | STM · DAVIS17 validation | J&F: 56.8 -> 61.2 | Improved |
| add cyclic training when reference is only the first frame (ablation) | STM · DAVIS16 validation | J&F: 81.2 -> 81.2 | No clear effect |
| add cyclic training with memory (MEM) reference (ablation) | STM · DAVIS17 validation | J&F: 69.7 -> 71.7 | Improved |
| add cyclic training to KMN (ablation) | KMN · DAVIS17 validation | J&F: 69.9 -> 70.7 | Improved |
| add gradient correction to cyclic KMN (ablation) | KMN-cycle · DAVIS17 validation | J&F: 70.7 -> 72.2 | Improved |
| add gradient correction to cyclic GCM (ablation) | GCM-cycle · DAVIS17 validation | J&F: 70.4 -> 70.7 | No clear effect |
| add cyclic training to AOT-T (ablation) | AOT-T · DAVIS17 validation | J&F: 79.2 -> 80.2 | Improved |
| add gradient correction on box-template noisy reference masks (ablation) | STM-cycle with MEM · DAVIS17 validation | J&F: 64.5 -> 71.8 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- AGSS-VOS — AGSS-VOS: Attention Guided Single-Shot Video Object Segmentation (ICCV 2019) · better on 2 of 2 shared results
- FEELVOS — FEELVOS: Fast End-To-End Embedding Learning for Video Object Segmentation (CVPR 2019) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Delving into the Cyclic Mechanism in Semi-supervised Video Object Segmentation (NeurIPS 2020 · similarity 0.96)
- Learning Position and Target Consistency for Memory-Based Video Object Segmentation (CVPR 2021 · similarity 0.91)
- Learning Dynamic Network Using a Reuse Gate Function in Semi-Supervised Video Object Segmentation (CVPR 2021 · similarity 0.90)
- A Transductive Approach for Video Object Segmentation (CVPR 2020 · similarity 0.89)
- From ViT Features to Training-free Video Object Segmentation via Streaming-data Mixture Models (NeurIPS 2023 · similarity 0.89)
- Accelerating Video Object Segmentation With Compressed Video (CVPR 2022 · similarity 0.89)
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.
