Research map / Video / Action recognition
SkillSpotter: Pose-Aware Multi-View Skilled Action Detection and Grading in Ego-Exo Videos
Björn Braun, Christian Holz
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 |
|---|---|---|---|
| add adaptive temporal suppression (vs No NMS) (ablation) | ActionFormer · Ego-Exo4D (Ego) | balanced accuracy (%): 54.81 -> 59.74 | Improved |
| add adaptive temporal suppression (vs No NMS) (ablation) | ActionFormer · Ego-Exo4D (Ego) | mAP_S: 13.96 -> 18.82 | Improved |
| replace Soft NMS with no suppression (ablation) | ActionFormer · Ego-Exo4D (Ego) | mAP_A / BA: 16.85 -> 22.50 mAP_A; 55.34 -> 54.81 BA | Mixed |
| add gated 3D body pose fusion (ablation) | ActionFormer + adaptive suppression · Ego-Exo4D (Ego) | mAP_S: 18.82 -> 21.82 | Improved |
| add gated 3D body pose fusion (ablation) | ActionFormer + adaptive suppression · Ego-Exo4D (Exos) | balanced accuracy (%): 58.29 -> 60.59 | Improved |
| add hand pose to gated pose fusion (ablation) | SkillSpotter · Ego-Exo4D (Ego) | mAP_S / BA: 21.82 -> 20.92 mAP_S; 60.40 -> 59.56 BA | Hurt |
| add bidirectional cross-view attention to concatenated Ego+Exos (ablation) | SkillSpotter · Ego-Exo4D (Ego+Exos) | balanced accuracy (%): 45.77 -> 60.39 | Improved |
| add cross-view attention to concatenated Ego+Exos (ablation) | SkillSpotter · Ego-Exo4D (Ego+Exos) | mAP_S: 21.99 -> 21.34 | No clear effect |
| naively concatenate ego and exo views | ActionFormer · Ego-Exo4D | balanced accuracy (%): 55.99 -> 50.34 | Hurt |
| train pose on ground truth instead of predicted pose (ablation) | SkillSpotter · Ego-Exo4D (Ego) | mAP_S: 19.96 -> 21.82 | Improved |
| add adaptive suppression and pose fusion to DyFADet (ablation) | DyFADet · Ego-Exo4D (Exos) | balanced accuracy (%): 49.52 -> 41.83 | Hurt |
| add adaptive suppression and pose fusion to TriDet (ablation) | TriDet · Ego-Exo4D (Ego) | mAP_S: 10.35 -> 20.47 | 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.
- ActionFormer — ActionFormer: Localizing Moments of Actions with Transformers (ECCV 2022) · better on 7 of 7 shared results
- TriDet — TriDet: Temporal Action Detection With Relative Boundary Modeling (CVPR 2023) · 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
- Ego-Exo4D: Understanding Skilled Human Activity from First- and Third-Person Perspectives (CVPR 2024 · similarity 0.91)
- ExpertAF: Expert Actionable Feedback from Video (CVPR 2025 · similarity 0.87)
- EgoSim: An Egocentric Multi-view Simulator and Real Dataset for Body-worn Cameras during Motion and Activity (NeurIPS 2024 · similarity 0.87)
- Test-Time Adaptation for Combating Missing Modalities in Egocentric Videos (ICLR 2025 · similarity 0.87)
- Understanding Multi-Task Activities from Single-Task Videos (CVPR 2025 · similarity 0.87)
- Bootstrap Your Own Views: Masked Ego-Exo Modeling for Fine-grained View-invariant Video Representations (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.
