Research map / Video / Action recognition
NurViD: A Large Expert-Level Video Database for Nursing Procedure Activity Understanding
Ming Hu, Lin Wang, Siyuan Yan, Don Ma, Qingli Ren, Peng Xia, Wei Feng, Peibo Duan, Lie Ju, Zongyuan Ge
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 |
|---|---|---|---|
| Initialize C3D from Kinetics-400 pretraining | C3D · procedure classification, untrimmed (Many) | per-class top-1 accuracy: 10.7 -> 21.5 | Improved |
| Initialize C3D from Kinetics-400 pretraining | C3D · procedure classification, untrimmed (Medium) | per-class top-1 accuracy: 5.1 -> 11.3 | Improved |
| Initialize SlowFast from Kinetics-400 pretraining | SlowFast · procedure classification, untrimmed (All) | per-class top-1 accuracy: 7.4 -> 13.5 | Improved |
| Initialize I3D from Kinetics-400 pretraining | I3D · procedure classification, untrimmed (All) | per-class top-1 accuracy: 8.7 -> 13.1 | Improved |
| Initialize C3D from Kinetics-400 pretraining | C3D · trimmed procedure classification (Many) | per-class top-1 accuracy: 70.1 -> 73.2 | Improved |
| Initialize C3D from Kinetics-400 pretraining | C3D · trimmed procedure classification (Medium) | per-class top-1 accuracy: 48.8 -> 60.0 | Improved |
| Initialize C3D from Kinetics-400 pretraining | C3D · trimmed procedure classification (Few) | per-class top-1 accuracy: 33.0 -> 39.6 | Improved |
| Initialize C3D from Kinetics-400 pretraining | C3D · trimmed action classification (All) | per-class top-1 accuracy: 15.9 -> 22.8 | Improved |
| Initialize C3D from Kinetics-400 pretraining | C3D · trimmed joint classification (All) | per-class top-1 accuracy: 7.7 -> 13.1 | Improved |
| Initialize SlowFast from Kinetics-400 pretraining | SlowFast · trimmed joint classification (All) | per-class top-1 accuracy: 7.5 -> 12.8 | Improved |
| Compare long-tail splits: few-shot classes vs many classes (C3D*, untrimmed) | C3D* · procedure classification, untrimmed | per-class top-1 accuracy: 21.5 -> 5.8 | Hurt |
| Use ActionFormer instead of TriDet for detection | ActionFormer · action detection | average mAP: 20.8 -> 23.9 | Improved |
Closest papers by meaning
- MedHorizon: Towards Long-context Medical Video Understanding in the Wild (NeurIPS 2026 · similarity 0.83)
- CaptainCook4D: A Dataset for Understanding Errors in Procedural Activities (NeurIPS 2024 · similarity 0.81)
- EgoExOR: An Ego-Exo-Centric Operating Room Dataset for Surgical Activity Understanding (NeurIPS 2025 · similarity 0.80)
- CLIMB: Data Foundations for Large Scale Multimodal Clinical Foundation Models (ICML 2025 · similarity 0.80)
- CPR-Coach: Recognizing Composite Error Actions based on Single-class Training (CVPR 2024 · similarity 0.80)
- CholecTrack20: A Multi-Perspective Tracking Dataset for Surgical Tools (CVPR 2025 · similarity 0.80)
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.
