Research map / Point clouds / Point cloud classification
Paving the Way for Point Cloud Video Representation Learning Using a PDE Model
Zhuoxu Huang, Zhenkun Fan, Jungong Han, Josef Kittler
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 MotionPDE regularizer | PSTNet · MSRAction-3D | accuracy: 91.20 -> 94.28 | Improved |
| add MotionPDE regularizer | PST-Transformer · MSRAction-3D | accuracy: 93.73 -> 96.63 | Improved |
| add MotionPDE regularizer | PSTNet · UTD-MHAD | accuracy: 93.49 -> 95.58 | Improved |
| add MotionPDE regularizer | PPTr · HOI4D action segmentation | accuracy: 77.4 -> 82.42 | Improved |
| add MotionPDE regularizer | PointLSTM · SHREC 2017 (no BBox) | accuracy: 92.50 -> 93.69 | Improved |
| add spectral method (trigonometric operators) to the PDE solver (ablation) | PSTNet + MotionPDE · MSRAction-3D | accuracy: 92.25 -> 94.28 | Improved |
| increase operators from 16 to 32 (ablation) | PSTNet + MotionPDE · MSRAction-3D | accuracy: 94.28 -> 91.91 | Hurt |
| reduce latent tokens from 4 to 2 (ablation) | PSTNet + MotionPDE · MSRAction-3D | accuracy: 94.28 -> 91.58 | Hurt |
| add max pooling on the spatial branch (ablation) | PSTNet + MotionPDE · MSRAction-3D | accuracy: 94.95 -> 94.28 | No clear effect |
| add max pooling on the spatial branch (ablation) | PST-Transformer + MotionPDE · MSRAction-3D | accuracy: 95.96 -> 96.63 | No clear effect |
| pre-train with MotionPDE contrastive loss then fine-tune | PST-Transformer · MSRAction-3D | accuracy: 94.33 -> 97.30 | Improved |
| pre-train on NTU60 (sparse stride) then fine-tune on MSR | PSTNet · MSRAction-3D | accuracy: 90.57 -> 90.23 | Hurt |
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.
- PSTNet — PSTNet: Point Spatio-Temporal Convolution on Point Cloud Sequences (ICLR 2021) · better on 5 of 5 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
- Adapting Pre-trained 3D Models for Point Cloud Video Understanding via Cross-frame Spatio-temporal Perception (CVPR 2025 · similarity 0.89)
- UST-SSM: Unified Spatio-Temporal State Space Models for Point Cloud Video Modeling (ICCV 2025 · similarity 0.88)
- Point Spatio-Temporal Transformer Networks for Point Cloud Video Modeling (TPAMI 2022 · similarity 0.88)
- PvNeXt: Rethinking Network Design and Temporal Motion for Point Cloud Video Recognition (ICLR 2025 · similarity 0.87)
- Sonata: Self-Supervised Learning of Reliable Point Representations (CVPR 2025 · similarity 0.86)
- 3DInAction: Understanding Human Actions in 3D Point Clouds (CVPR 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.
