Research map / Point clouds / Point cloud classification
Masked Motion Prediction with Semantic Contrast for Point Cloud Sequence Learning
yuehui han, Can Xu, Rui Xu, Jianjun Qian, Jin Xie
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 motion trajectory prediction and global semantic contrast to masked point cloud pretraining (M2PSC vs MaST-Pre) | P4Transformer · MSRAction-3D | accuracy (%): 91.29 -> 93.03 | Improved |
| M2PSC pretraining instead of MaST-Pre | PST-Transformer · MSRAction-3D | accuracy (%): 94.08 -> 94.84 | Improved |
| M2PSC pretraining instead of MaST-Pre | P4Transformer · NTU-RGBD (end-to-end) | accuracy (%): 90.8 -> 91.3 | Improved |
| M2PSC pretraining instead of MaST-Pre | P4Transformer · NTU-RGBD (50% semi-supervised) | accuracy (%): 87.8 -> 88.7 | Improved |
| M2PSC pretraining instead of MaST-Pre | P4Transformer · NvGesture (transfer, 50 epochs) | accuracy (%): 89.3 -> 89.6 | Improved |
| M2PSC pretraining instead of MaST-Pre | P4Transformer · SHREC'17 (transfer, 50 epochs) | accuracy (%): 92.4 -> 92.8 | Improved |
| use all three pretext tasks instead of a single task (ablation) | M2PSC · MSRAction-3D | accuracy (%): 89.73 -> 93.03 | Improved |
| add other pretext tasks to motion trajectory prediction (ablation) | M2PSC · MSRAction-3D | accuracy (%): 89.73 -> 92.67 / 92.83 | Improved |
| M2PSC (P4Transformer) vs supervised-only gesture models such as Kinet | P4Transformer + M2PSC · SHREC'17 (transfer, 50 epochs) | accuracy (%): 95.2 -> 92.8 | Hurt |
Closest papers by meaning
- Contrastive Predictive Autoencoders for Dynamic Point Cloud Self-Supervised Learning (AAAI 2023 · similarity 0.92)
- Masked Spatio-Temporal Structure Prediction for Self-supervised Learning on Point Cloud Videos (ICCV 2023 · similarity 0.90)
- PointCMP: Contrastive Mask Prediction for Self-Supervised Learning on Point Cloud Videos (CVPR 2023 · similarity 0.90)
- Self-Supervised Class-Agnostic Motion Prediction with Spatial and Temporal Consistency Regularizations (CVPR 2024 · similarity 0.85)
- PvNeXt: Rethinking Network Design and Temporal Motion for Point Cloud Video Recognition (ICLR 2025 · similarity 0.85)
- RI-MAE: Rotation-Invariant Masked AutoEncoders for Self-Supervised Point Cloud Representation Learning (AAAI 2025 · 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.
