Research map / Detection and segmentation / Open-set and few-shot detection
Segmenting Moving Objects via an Object-Centric Layered Representation
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 instance normalisation to CNN encoder (ablation) | OCLR · DAVIS2017-motion | J mean: 48.7 -> 54.5 | Helped |
| add amodal mask supervision (vs modal only) (ablation) | OCLR · DAVIS2017-motion | J mean: 50.5 -> 54.5 | Helped |
| add amodal mask supervision (vs modal only) (ablation) | OCLR · Syn-Val amodal | J mean: 76.9 -> 84.7 | Helped |
| longer temporal input (30 vs 15 frames) (ablation) | OCLR · DAVIS2017-motion | J mean: 53.5 -> 54.5 | Helped |
| add boundary loss (weight 0.2) (ablation) | OCLR · DAVIS2017-motion | J mean: 54.1 -> 54.5 | Helped |
| add boundary loss (weight 0.2) (ablation) | OCLR · Syn-Val modal | J mean: 80.9 -> 85.6 | Helped |
| use Hungarian matching for layer assignment (ablation) | OCLR · DAVIS2017-motion | J mean: 44.2 -> 54.5 | Helped |
| train on RAFT flow instead of ground-truth flow (ablation) | OCLR synthetic pipeline · DAVIS2016 | J mean: 67.4 -> 68.6 | Helped |
| add real video backgrounds to synthetic data (ablation) | OCLR synthetic pipeline · DAVIS2016 | J mean: 68.6 -> 72.0 | Helped |
| add RGB-based test-time adaptation | OCLR · DAVIS2016 | J mean: 72.1 -> 80.9 | Helped |
| add RGB-based test-time adaptation | OCLR · MoCA | success rate: 0.599 -> 0.559 | Hurt |
| add RGB-based test-time adaptation | OCLR · DAVIS2017-motion | J&F: 55.1 -> 64.4 | Helped |
Closest papers by meaning
- Using Diffusion Priors for Video Amodal Segmentation (CVPR 2025 · similarity 0.89)
- GMOS: Grounding Moving Object Segmentation in 3D Space and Time (NeurIPS 2026 · similarity 0.89)
- Segment Any Motion in Videos (CVPR 2025 · similarity 0.88)
- M^3-VOS: Multi-Phase, Multi-Transition, and Multi-Scenery Video Object Segmentation (CVPR 2025 · similarity 0.88)
- Learning Segmentation from Point Trajectories (NeurIPS 2024 · similarity 0.88)
- SMITE: Segment Me In TimE (ICLR 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.