Research map / Video / Video segmentation and tracking
Temporal Transductive Inference for Few-Shot Video Object Segmentation
Mennatullah Siam
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 global video-level loss (ablation) | RePRI baseline · YouTube-VIS FS-VOS | mVC3: 66.3 -> 67.3 | Improved |
| add keyframe refinement (ablation) | TTI with global loss · YouTube-VIS FS-VOS | mVC3: 67.3 -> 70.6 | Improved |
| add dense contrastive learning (ablation) | TTI with global loss and keyframe refinement · YouTube-VIS FS-VOS | mVC3: 70.6 -> 72.1 | Improved |
| add dense contrastive learning (ablation) | TTI with global loss and keyframe refinement · YouTube-VIS FS-VOS fold 1 | mVC3: 59.6 -> 57.7 | Hurt |
| learn per-frame classifiers with temporal regularization instead of single set of weights over all frames (ablation) | Naive Temporal RePRI · YouTube-VIS FS-VOS | mIoU: 51.0 -> 60.5 | Improved |
| add temporal transductive inference to RePRI | RePRI · YouTube-VIS FS-VOS | mIoU: 59.5 -> 60.5 | Improved |
| add temporal transductive inference to RePRI | RePRI · YouTube-VIS FS-VOS | mVC3: 66.3 -> 72.1 | Improved |
| add temporal transductive inference to RePRI | RePRI · YouTube-VIS FS-VOS fold 4 | mIoU: 64.2 -> 62.4 | Hurt |
| add temporal transductive inference to RePRI | RePRI · MiniVSPW | mIoU: 27.1 -> 29.3 | Improved |
| add temporal transductive inference to RePRI | RePRI · MiniVSPW | mVC3: 22.9 -> 26.9 | Improved |
| add temporal transductive inference to RePRI | RePRI · YouTube-VIS FS-VOS | boundary accuracy F: 43.5 -> 45.0 | Mixed |
| operate on final linear classifiers instead of fine-tuning backbone online | TTI vs DANet · YouTube-VIS FS-VOS | runtime per video: 20 s -> 3 s | Improved |
Closest papers by meaning
- Learning What to Learn for Video Object Segmentation (ECCV 2020 · similarity 0.91)
- Few-Shot Segmentation Without Meta-Learning: A Good Transductive Inference Is All You Need? (CVPR 2021 · similarity 0.90)
- A Transductive Approach for Video Object Segmentation (CVPR 2020 · similarity 0.90)
- Video Object Segmentation without Temporal Information (TPAMI 2018 · similarity 0.90)
- Efficient Semantic Video Segmentation with Per-frame Inference (ECCV 2020 · similarity 0.89)
- Less than Few: Self-Shot Video Instance Segmentation (ECCV 2022 · similarity 0.88)
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.
