Research map / Point clouds / Point cloud classification
MS^2Gait: A Multi-Scale Spatio-Temporal Fusion Network for LiDAR-based Gait Recognition
Shenyin Xu, Yishan Wang, Xinyu Li, Rui Liu, Zhongyuan Wang, Xin Tian
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 Inter-Set Mixing to Intra-Set Mixing (ablation) | MS2Gait HSFE · SUSTech1K | OA Rank-1: 84.2 -> 88.3 | Improved |
| add Intra-Layer Mixing to Intra-Set Mixing (ablation) | MS2Gait HSFE · SUSTech1K | OA Rank-1: 84.2 -> 87.5 | Improved |
| add Inter-Layer Mixing to Intra-Set Mixing (ablation) | MS2Gait HSFE · SUSTech1K | OA Rank-1: 84.2 -> 85.9 | Improved |
| use all four mixing strategies instead of Intra-Set only (ablation) | MS2Gait HSFE · SUSTech1K | OA Rank-1: 84.2 -> 90.6 | Improved |
| add GMFF multi-resolution fusion (ablation) | MS2Gait · SUSTech1K | OA Rank-1: 90.6 -> 92.0 | Improved |
| replace temporal pooling with Transformer encoder (ablation) | MS2Gait · SUSTech1K | OA Rank-1: 92.0 -> 92.2 | No clear effect |
| add STET key frame selection and similarity aggregation on top of Transformer (ablation) | MS2Gait · SUSTech1K | OA Rank-1: 92.2 -> 93.5 | Improved |
| add MS2Gait components to PointNet++ baseline | PointNet++ · FreeGait | Rank-1: 59.3 -> 83.1 | Improved |
| add MS2Gait components to PointNet++ baseline | PointNet++ · SUSTech1K Bag subset | Rank-1: 78.7 -> 94.7 | Improved |
| MS2Gait versus LidarGait++ on clothing subset | MS2Gait · SUSTech1K Clothing subset | Rank-1: 79.7 -> 79.7 | No clear effect |
| MS2Gait versus LidarGait++ on occlusion subset | MS2Gait · SUSTech1K Occlusion subset | Rank-1: 96.6 -> 95.9 | Hurt |
| add MS2Gait modules (spatial interaction and STET) to PointNet++ | PointNet++ · SUSTech1K inference | total inference time (ms): 149.38 -> 157.33 | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- LidarGait++ — LidarGait: Benchmarking 3D Gait Recognition With Point Clouds (CVPR 2023) · better on 8 of 11 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
- LidarGait: Benchmarking 3D Gait Recognition With Point Clouds (CVPR 2023 · similarity 0.92)
- LidarGait++: Learning Local Features and Size Awareness from LiDAR Point Clouds for 3D Gait Recognition (CVPR 2025 · similarity 0.91)
- Walking Further: Semantic-Aware Multimodal Gait Recognition Under Long-Range Conditions (AAAI 2026 · similarity 0.90)
- Camera-LiDAR Cross-modality Gait Recognition (ECCV 2024 · similarity 0.87)
- MMGait: Towards Multi-Modal Gait Recognition (CVPR 2026 · similarity 0.87)
- HybridGait: A Benchmark for Spatial-Temporal Cloth-Changing Gait Recognition with Hybrid Explorations (AAAI 2024 · 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.
