Research map / Detection and segmentation / Open-set and few-shot detection
MOVE: Unsupervised Movable Object Segmentation and Detection
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 |
|---|---|---|---|
| use no shift (ablation) | MOVE · HKU-IS val | IoU: 0.819 -> 0.000 | Hurt |
| shift range 1/16 instead of 2/16 (ablation) | MOVE · HKU-IS val | IoU: 0.819 -> 0.751 | Hurt |
| shift range 3/16 instead of 2/16 (ablation) | MOVE · HKU-IS val | IoU: 0.819 -> 0.799 | Hurt |
| shift range 4/16 instead of 2/16 (ablation) | MOVE · HKU-IS val | IoU: 0.819 -> 0.704 | Hurt |
| discriminator fake inputs only composed images (ablation) | MOVE · HKU-IS val | IoU: 0.819 -> 0.789 | Hurt |
| discriminator real inputs: x plus composed without shift (ablation) | MOVE · HKU-IS val | IoU: 0.819 -> 0.740 | Hurt |
| non-differentiable inpainter (ablation) | MOVE · HKU-IS val | IoU: 0.819 -> 0.314 | Hurt |
| MSE-trained MAE inpainter instead of GAN-trained (ablation) | MOVE · HKU-IS val | IoU: 0.819 -> 0.817 | No clear effect |
| MAE features instead of DINO (ablation) | MOVE · HKU-IS val | IoU: 0.819 -> 0.783 | Hurt |
| train on ImageNet100 instead of DUTS-TR (ablation) | MOVE · HKU-IS val | IoU: 0.819 -> 0.815 | No clear effect |
| add bilateral solver to MOVE masks | MOVE · DUTS-TE | IoU: .713 -> .687 | Hurt |
| train class-agnostic detector on Multi boxes instead of single box | MOVE + CAD · COCO20K | CorLoc: 69.1 -> 71.9 | Helped |
Closest papers by meaning
- Learning Features by Watching Objects Move (CVPR 2017 · similarity 0.89)
- Unsupervised Multi-Object Segmentation by Predicting Probable Motion Patterns (NeurIPS 2022 · similarity 0.89)
- GMOS: Grounding Moving Object Segmentation in 3D Space and Time (NeurIPS 2026 · similarity 0.86)
- Unsupervised Object Detection with Theoretical Guarantees (NeurIPS 2024 · similarity 0.86)
- Segment Any Motion in Videos (CVPR 2025 · similarity 0.86)
- Segment Anything without Supervision (NeurIPS 2024 · similarity 0.86)
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.