Research map / Detection and segmentation / Open-set and few-shot detection
MOVES: Manipulated Objects in Video Enable Segmentation
Richard E. L. Higgins, David F. Fouhey
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 |
|---|---|---|---|
| Use MOVES features for held-object segmentation instead of COHESIV | EPICK HOS · EPICK | Object mIoU (%): 19.5 →25.7 | Improved |
| Use MOVES features for Box2Seg instead of COHESIV | VISOR Box2Seg · VISOR held objects | mIoU: 8.9 →44.2 | Improved |
| Use MOVES features for Box2Seg instead of COHESIV | VISOR Box2Seg · VISOR non-held objects | mIoU: 7.5 →45.0 | Improved |
| Use MOVES feature distances to determine whether pixels belong to the same object instead of RGB distances | Object grouping · VISOR and FG | AUROC: 99.5 vs 81.5; 95.2 vs 78.4; 95.0 vs 78.6; 97.1 vs 75.8; 94.7 vs 73.3 | Improved |
| Train instance segmentation using MOVES Box2Seg masks rather than painstakingly annotated masks | VISOR HOS instance segmentation · VISOR | Mask AP@50 Held: 49.8 vs 25.5 | Hurt |
| Replace GMA-Flow with PWC-Net optical flow (ablation) | MOVES · VISOR Hand feature distance | AUROC: 99.5 -> 98.2 | No clear effect |
| Replace GMA-Flow with PWC-Net optical flow (ablation) | MOVES · VISOR Held feature distance | AUROC: 95.2 -> 94.6 | No clear effect |
| Replace GMA-Flow with PWC-Net optical flow (ablation) | MOVES · VISOR Non-held feature distance | AUROC: 95.0 -> 94.2 | No clear effect |
| Use connected components on above-mean optical-flow magnitude for grouping pseudolabels (ablation) | MOVES · VISOR Hand feature distance | AUROC: 99.5 -> 98.9 | No clear effect |
| Use MOVES with grouping loss on EISEN playroom rather than EISEN | Object segmentation · EISEN playroom | mIoU: 73.0 vs 71.8 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- COHESIV — COHESIV: Contrastive Object and Hand Embedding Segmentation In Video (NeurIPS 2021) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- COHESIV: Contrastive Object and Hand Embedding Segmentation In Video (NeurIPS 2021 · similarity 0.89)
- "Fine-Grained Egocentric Hand-Object Segmentation: Dataset, Model, and Applications" (ECCV 2022 · similarity 0.88)
- Analysis of Hand Segmentation in the Wild (CVPR 2018 · similarity 0.87)
- Human Hands As Probes for Interactive Object Understanding (CVPR 2022 · similarity 0.87)
- HOIST-Former: Hand-held Objects Identification Segmentation and Tracking in the Wild (CVPR 2024 · similarity 0.85)
- Modeling Fine-Grained Hand-Object Dynamics for Egocentric Video Representation Learning (ICLR 2025 · similarity 0.85)
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.
