Research map / Video / Video segmentation and tracking
HODOR: High-Level Object Descriptors for Object Re-Segmentation in Video Learned 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 affine augmentation across the repeated static-image sequence | HODOR · DAVIS 2017 val | J&F: 69.4 -> 77.5 | Helped |
| Add cyclic-consistency fine-tuning on unlabeled video frames, using one annotated frame per video | HODOR · DAVIS 2017 val | J&F: 77.5 -> 80.6 | Helped |
| Train with temporally dense video annotations | HODOR · DAVIS 2017 val | J&F: 77.5 -> 81.3 | Helped |
| Train on dense video annotations instead of relying on the HODOR bottleneck | HODOR vs. STCN · DAVIS 2017 val | J&F: 81.3 vs. 85.4 | Hurt |
| Add repeated-image sequence training without augmentation | HODOR · DAVIS 2017 val | J&F: 61.6 -> 69.4 | Helped |
| Use multi-instance processing rather than one foreground instance per pass (ablation) | HODOR · DAVIS 2017 val | J&F: 71.5 -> 77.5 | Helped |
| Add learned soft attention masking in the encoder (ablation) | HODOR · DAVIS 2017 val | J&F: 74.4 -> 77.5 | Helped |
| Use learned soft attention masking instead of hard attention masking (ablation) | HODOR · DAVIS 2017 val | J&F: 74.5 -> 77.5 | Helped |
| Add encoder layers (five instead of zero) (ablation) | HODOR · DAVIS 2017 val | J&F: 72.8 -> 77.5 | Helped |
| Add decoder layers (five instead of zero) (ablation) | HODOR · DAVIS 2017 val | J&F: 74.4 -> 77.5 | Helped |
| Add the encoder and decoder rather than using the barebones network (ablation) | HODOR · DAVIS 2017 val | J&F: 70.6 -> 77.5 | Helped |
| Use deformable rather than regular convolution in the decoder (ablation) | HODOR · DAVIS 2017 val | J&F: 75.1 -> 77.5 | Helped |
Closest papers by meaning
- Breaking the "Object" in Video Object Segmentation (CVPR 2023 · similarity 0.91)
- Two-Shot Video Object Segmentation (CVPR 2023 · similarity 0.89)
- Point-VOS: Pointing Up Video Object Segmentation (CVPR 2024 · similarity 0.88)
- Look Before You Match: Instance Understanding Matters in Video Object Segmentation (CVPR 2023 · similarity 0.88)
- Putting the Object Back into Video Object Segmentation (CVPR 2024 · similarity 0.88)
- M^3-VOS: Multi-Phase, Multi-Transition, and Multi-Scenery Video Object Segmentation (CVPR 2025 · 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.