Research map / Point clouds / Point cloud classification
An Efficient PointLSTM for Point Clouds Based Gesture Recognition
Yuecong Min, Yanxiao Zhang, Xiujuan Chai, Xilin Chen
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 |
|---|---|---|---|
| use PointLSTM-early instead of the baseline | point-cloud gesture recognition · NVGesture | accuracy: 85.9(±0.5) -> 87.9(±0.7) | Improved |
| use PointLSTM-middle instead of the baseline | point-cloud gesture recognition · SHREC’17, 28 gestures | accuracy: 87.6 -> 94.7 | Improved |
| use PointLSTM-late instead of the baseline | point-cloud action recognition · MSR Action3D | accuracy: 87.62±1.48 -> 92.29±3.09 | Improved |
| replace the baseline with a frame-level LSTM (ablation) | point-cloud gesture recognition · NVGesture | accuracy: 85.9(±0.5) -> 82.8(±0.8) | Hurt |
| replace stage-1 with PointLSTM on raw points (ablation) | point-cloud gesture recognition · NVGesture | accuracy: 85.9(±0.5) -> 82.5(±1.3) | Hurt |
| replace stage-2 with PointLSTM using k=1 (ablation) | point-cloud gesture recognition · NVGesture | accuracy: 85.9(±0.5) -> 87.9(±0.7) | Improved |
| replace stage-2 with PointLSTM using k=4 instead of k=1 (ablation) | point-cloud gesture recognition · NVGesture | accuracy: 87.9(±0.7) -> 86.4(±0.5) | Hurt |
| replace stage-3 with PointLSTM using k=16 (ablation) | point-cloud gesture recognition · NVGesture | accuracy: 85.9(±0.5) -> 86.9(±0.6) | Improved |
| use point-shared states instead of point-independent states (ablation) | PointLSTM gesture recognition · SHREC’17, 28 gestures | accuracy: 94.7(±0.1) -> 93.1(±0.2) | Hurt |
| use point-shared states instead of the full PointLSTM (ablation) | PointLSTM gesture recognition · NVGesture | accuracy: PointLSTM-middle k=1: 85.4(±0.6) -> PSS 87.3(±0.4); PointLSTM-middle k=16: 86.9(±0.6) -> PSS 87.3(±0.4) | Mixed |
| use centroid-aligned grouping instead of direct grouping (ablation) | PointLSTM-middle · NVGesture | accuracy | Improved |
| use centroid-aligned grouping instead of direct grouping (ablation) | PointLSTM-middle · SHREC’17 | accuracy | Hurt |
Closest papers by meaning
- Adapting Pre-trained 3D Models for Point Cloud Video Understanding via Cross-frame Spatio-temporal Perception (CVPR 2025 · similarity 0.84)
- PointRWKV: Efficient RWKV-Like Model for Hierarchical Point Cloud Learning (AAAI 2025 · similarity 0.84)
- PSTNet: Point Spatio-Temporal Convolution on Point Cloud Sequences (ICLR 2021 · similarity 0.83)
- 3DInAction: Understanding Human Actions in 3D Point Clouds (CVPR 2024 · similarity 0.83)
- PointMamba: A Simple State Space Model for Point Cloud Analysis (NeurIPS 2024 · similarity 0.83)
- Point-Cache: Test-time Dynamic and Hierarchical Cache for Robust and Generalizable Point Cloud Analysis (CVPR 2025 · similarity 0.82)
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.
