Research map / Point clouds / Point cloud classification
Tri-Efficient Transfer Learning for Point Cloud Videos
Yiding Sun, Dongxu Zhang, Jihua Zhu, Haozhe Cheng, Zhengqiao Li, Pengcheng Li, Chaowei Fang, Yonghao Dong, Lin Chen
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 pseudo-motion consistency objective (ablation) | GMD Net pretraining · MSR-Action3D (pretraining ablation, no fine-tuning) | accuracy (%): 83.50 -> 93.03 | Improved |
| add cross-modal (image and text) contrastive loss (ablation) | GMD Net pretraining · MSR-Action3D (pretraining ablation, no fine-tuning) | accuracy (%): 93.03 -> 94.42 | Improved |
| add KL divergence motion distribution loss (ablation) | GMD Net pretraining · MSR-Action3D (pretraining ablation, no fine-tuning) | accuracy (%): 94.42 -> 95.42 | Improved |
| add additive adapter (PointCSA style) to pretrained backbone (ablation) | PoinTriE · MSR-Action3D (fine-tuning ablation) | accuracy (%): 95.42 -> 96.16 | Improved |
| use side tuning instead of additive adapter (ablation) | PoinTriE · MSR-Action3D (fine-tuning ablation) | accuracy (%): 96.16 -> 96.86 | Improved |
| add gradient flow masking to side network (ablation) | PoinTriE · MSR-Action3D (fine-tuning ablation) | accuracy (%): 96.86 -> 97.21 | Improved |
| PoinTriE instead of PointATA | fine-tuning paradigm · MSR-Action3D | avg accuracy (%): 93.32 -> 94.37 | Improved |
| PoinTriE instead of PointCSA, memory | fine-tuning paradigm · MSR-Action3D | memory: 26.4 -> 9.8 | Improved |
| PoinTriE instead of PointATA | fine-tuning paradigm · SHREC'17 | accuracy (%): 95.5 -> 96.5 | Improved |
| PoinTriE instead of PointATA with 3 frames | fine-tuning paradigm · Synthia 4D | mIoU (%): 84.06 -> 84.11 | No clear effect |
| PoinTriE instead of PointATA at 8 frames | fine-tuning paradigm · MSR-Action3D (8 frames) | accuracy (%): 91.28 -> 91.28 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- PointCMP — PointCMP: Contrastive Mask Prediction for Self-Supervised Learning on Point Cloud Videos (CVPR 2023) · better on 2 of 2 shared results
- MAMBA4D — Mamba4D: Efficient 4D Point Cloud Video Understanding with Disentangled Spatial-Temporal State Space Models (CVPR 2025) · better on 1 of 1 shared result
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)
- Point-PEFT: Parameter-Efficient Fine-Tuning for 3D Pre-trained Models (AAAI 2024 · similarity 0.89)
- 3DInAction: Understanding Human Actions in 3D Point Clouds (CVPR 2024 · similarity 0.88)
- On Geometry-Enhanced Parameter-Efficient Fine-Tuning for 3D Scene Segmentation (NeurIPS 2025 · similarity 0.88)
- PvNeXt: Rethinking Network Design and Temporal Motion for Point Cloud Video Recognition (ICLR 2025 · similarity 0.88)
- Sonata: Self-Supervised Learning of Reliable Point Representations (CVPR 2025 · similarity 0.87)
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.
