Research map / Video / Action recognition
Assembly101: A Large-Scale Multi-View Video Dataset for Understanding Procedural Activities
Check 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 fixed-camera rather than egocentric views | TSM recognition on Assembly101 · Assembly101 | Top-1 action accuracy: 23.0 -> 39.2 | Helped |
| Train on both fixed and egocentric views rather than egocentric views alone | TSM recognition on Assembly101 · Assembly101 | Top-1 action accuracy: 23.0 -> 34.0 | Helped |
| Use fixed-camera rather than egocentric views | TempAgg anticipation on Assembly101 · Assembly101 | Top-5 action recall: 5.5 -> 10.4 | Helped |
| Use fixed-camera rather than egocentric views | C2F-TCN segmentation on Assembly101 · Assembly101 | MoF: 34.8 -> 41.3 | Helped |
| Replace MS-TCN++ with C2F-TCN | Temporal segmentation on Assembly101 · Assembly101 | F1@10: 31.6 -> 33.3 | Helped |
| Test on unseen rather than seen toys | C2F-TCN segmentation, disassembly · Assembly101 | F1@10: 35.8 -> 31.9 | Hurt |
| Extend pose action boundaries by 0.5 seconds (ablation) | 2s-AGCN pose recognition · Assembly101 | Top-1 action accuracy: 22.2 -> 26.7 | Helped |
| Add ground-truth object labels as frame-level features (ablation) | MS-G3D with context pose recognition · Assembly101 | Top-1 action accuracy: 28.7 -> 62.0 | Helped |
| Pretrain TSM features on Assembly101 instead of EPIC-KITCHENS-100 | TempAgg recognition · Assembly101 | Top-1 action accuracy: 17.3 -> 40.5 | Helped |
| Use full-segment recognition rather than half-segment early prediction | TempAgg mistake detection with TSM features · Assembly101 | Mistake-class recall: 35.0 -> 46.6 | Helped |
| Use full-segment recognition rather than half-segment early prediction | TempAgg mistake detection with TSM features · Assembly101 | Correction-class recall: 26.4 -> 29.6 | Helped |
Closest papers by meaning
- Ego-Exo4D: Understanding Skilled Human Activity from First- and Third-Person Perspectives (CVPR 2024 · similarity 0.87)
- EgoExoLearn: A Dataset for Bridging Asynchronous Ego- and Exo-centric View of Procedural Activities in Real World (CVPR 2024 · similarity 0.86)
- CaptainCook4D: A Dataset for Understanding Errors in Procedural Activities (NeurIPS 2024 · similarity 0.86)
- Error Detection in Egocentric Procedural Task Videos (CVPR 2024 · similarity 0.86)
- GigaHands: A Massive Annotated Dataset of Bimanual Hand Activities (CVPR 2025 · similarity 0.85)
- IndEgo: A Dataset of Industrial Scenarios and Collaborative Work for Egocentric Assistants (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.