Research map / Video / Video generation
ExpertEdit: Learning Skill-Aware Motion Editing from Expert Videos
Arjun Somayazulu, Kristen Grauman
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 expert-manifold kinematic-masked infilling instead of text-driven supervised editor | SimMotionEdit · Ego-Exo4D Rev. Layup | FID improvement F (%): 4.88 -> 12.08 | Improved |
| use ExpertEdit instead of SimMotionEdit on penalty kick | SimMotionEdit · Ego-Exo4D Pen. Kick | Pose improvement P (%): 3.20 -> 6.03 | Improved |
| use ExpertEdit instead of SimMotionEdit on Rev. Punch | SimMotionEdit · Kyokushin Karate Rev. Punch | Pose improvement P (%): 2.20 -> 2.07 | Hurt |
| use ExpertEdit instead of SimMotionEdit on High RH | SimMotionEdit · Kyokushin Karate High RH | Pose improvement P (%): 3.30 -> 2.96 | Hurt |
| use ExpertEdit instead of SimMotionEdit on karate Front kick | SimMotionEdit · Kyokushin Karate Front | FID improvement F (%): 5.05 -> 9.73 | Improved |
| prompt asking for smoother, controlled motion instead of expert-like (ablation) | SimMotionEdit · Ego-Exo4D Rev. Layup | FID improvement F (%): 0.81 -> 2.90 | Improved |
| add technique name to smooth prompt (ablation) | SimMotionEdit · Ego-Exo4D Pen. Kick | Pose improvement P (%): 1.74 -> 3.20 | Improved |
| increase training expert data from 30% to 100% (ablation) | ExpertEdit · Average over techniques | P and F | Improved |
| halve mask span fraction alpha 0.3 -> 0.15 (ablation) | ExpertEdit · Average over techniques | Pose improvement P (%): 4.02 -> 4.09; F 7.01 -> 6.10 | Mixed |
| increase mask span fraction alpha 0.3 -> 0.45 (ablation) | ExpertEdit · Average over techniques | FID improvement F (%): 7.01 -> 6.41 | Hurt |
| ExpertEdit vs SimMotionEdit in human preference | ExpertEdit · Human preference (basketball+soccer) | preferred %: 43.8% vs 25.9% | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- SimMotionEdit — SimMotionEdit: Text-Based Human Motion Editing with Motion Similarity Prediction (CVPR 2025) · better on 14 of 16 shared results
- FLAME — FLAME: Free-Form Language-Based Motion Synthesis & Editing (AAAI 2023) · better on 4 of 4 shared results
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
- ExpertAF: Expert Actionable Feedback from Video (CVPR 2025 · similarity 0.91)
- Learning Skill-Attributes for Transferable Assessment in Video (NeurIPS 2025 · similarity 0.87)
- CigTime: Corrective Instruction Generation Through Inverse Motion Editing (NeurIPS 2024 · similarity 0.87)
- Dynamic Motion Blending for Versatile Motion Editing (CVPR 2025 · similarity 0.86)
- Learning Action and Reasoning-Centric Image Editing from Videos and Simulation (NeurIPS 2024 · similarity 0.85)
- ExAct: A Video-Language Benchmark for Expert Action Analysis (NeurIPS 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.
