Research map / Video / Video-language understanding
Decoding Children’s Gait Behavior
Yifan Shen, Boyi Li, Meihuan Huang, Yuanzhe Liu, Xu Cao, Jinyang Jin, Zhengyuan Li, Anglin Liu, Junho Kim, Jingyuan Zhu, Lan Fangzhou, Jianguo Cao and 3 more
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 |
|---|---|---|---|
| Fine-tune Qwen3-VL-4B on CGV | Qwen3-VL-4B · CGV test | average accuracy: L 51 -> 53, R 52 -> 53 | No clear effect |
| Use video model (VideoMAE v2) fine-tuned on CGV instead of gait models such as BiggerGait | VideoMAE v2 · CGV test | average accuracy: L 54 -> 69, R 53 -> 72 | Improved |
| Add token-level kinematic prompting and mask-guided patch pruning to VideoMAE v2 | ChildGait-Video · CGV test | average accuracy: L 69 -> 84, R 72 -> 84 | Improved |
| Add token-level kinematic prompting (TKP) (ablation) | VideoMAE v2 · CGV test | average accuracy: L 69 -> 72, R 72 -> 74 | Improved |
| Add mask-guided patch pruning (MPP) (ablation) | VideoMAE v2 · CGV test | average accuracy: L 69 -> 78, R 72 -> 79 | Improved |
| Use bounding-box mask instead of mask-guided pruning (ablation) | VideoMAE v2 + MPP · CGV test | average accuracy: L 78 -> 75, R 79 -> 77 (MPP-only vs bbox) | Hurt |
| Replace MPP with random mask (ablation) | VideoMAE v2 · CGV test | average accuracy: L 69 -> 68, R 72 -> 70 | Hurt |
| Increase input frames from 8 to 16 (ablation) | ChildGait-Video · CGV item IC | accuracy: 74.1 -> 87.0 | Improved |
| Increase input frames from 16 to 32 (ablation) | ChildGait-Video · CGV item IC | accuracy: 87.0 -> 90.7 | Improved |
| Use zero-shot MLLMs for gait scoring | Gemini 3 Pro, GPT-5.2, Qwen3-VL etc. · CGV test | average accuracy: 50-59% vs ~50% chance | Hurt |
| Use adult-pretrained gait models fine-tuned on CGV | BiggerGait, SwinGait, GaitBase etc. · CGV test | average accuracy: 45%-56% | Hurt |
Closest papers by meaning
- Care-PD: A Multi-Site Anonymized Clinical Dataset for Parkinson’s Disease Gait Assessment (NeurIPS 2025 · similarity 0.85)
- Cross-Covariate Gait Recognition: A Benchmark (AAAI 2024 · similarity 0.85)
- On Denoising Walking Videos for Gait Recognition (CVPR 2025 · similarity 0.84)
- Exploring More from Multiple Gait Modalities for Human Identification (AAAI 2025 · similarity 0.84)
- Vocabulary-Guided Gait Recognition (NeurIPS 2025 · similarity 0.84)
- Learning Visual Prompt for Gait Recognition (CVPR 2024 · similarity 0.84)
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.
