Research map / Detection and segmentation / Open-set and few-shot detection
Learning Video Object Segmentation From Static Images
Check 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 optical-flow guidance (ablation) | MaskTrack · DAVIS | mIoU: 74.8 → 78.4 | Helped |
| Add CRF post-processing to MaskTrack + Flow (ablation) | MaskTrack · DAVIS | mIoU: 78.4 -> 80.3 | Helped |
| Add optical-flow guidance | MaskTrack · DAVIS, YouTubeObjects, and SegTrack-v2 | mIoU: 1∼4 % | Mixed |
| Add online fine-tuning on annotated frame(s) (ablation) | MaskTrack · DAVIS | mIoU: 69.9 -> 74.8 | Helped |
| Add offline training (ablation) | MaskTrack · DAVIS | mIoU: 57.6 -> 74.8 | Helped |
| Reduce static-image offline training data from 11k to 5k (ablation) | MaskTrack · DAVIS | mIoU: 74.8 -> 73.2 | Hurt |
| Train offline on video data instead of static images (ablation) | MaskTrack · DAVIS | mIoU: 74.8 -> 72.0 | Hurt |
| Add mask dilation during training-sample generation (ablation) | MaskTrack · DAVIS | mIoU: 72.4 -> 74.8 | Helped |
| Add mask deformation during training-sample generation (ablation) | MaskTrack · DAVIS | mIoU: 17.1 -> 74.8 | Helped |
| Add non-rigid mask deformation (ablation) | MaskTrack · DAVIS | mIoU: 73.3 -> 74.8 | Helped |
| Use segment masks instead of boxes as the extra input channel (ablation) | MaskTrack · DAVIS | mIoU: 69.6 -> 74.8 | Helped |
| Add the mask input channel (ablation) | MaskTrack · DAVIS | mIoU: 72.5 -> 74.8 | Helped |
Closest papers by meaning
- From ViT Features to Training-free Video Object Segmentation via Streaming-data Mixture Models (NeurIPS 2023 · similarity 0.89)
- MaskRNN: Instance Level Video Object Segmentation (NeurIPS 2017 · similarity 0.88)
- Efficient Video Object Segmentation via Network Modulation (CVPR 2018 · similarity 0.88)
- Breaking the "Object" in Video Object Segmentation (CVPR 2023 · similarity 0.88)
- A Generative Appearance Model for End-To-End Video Object Segmentation (CVPR 2019 · similarity 0.88)
- Putting the Object Back into Video Object Segmentation (CVPR 2024 · similarity 0.87)
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.