Research map / Video / Action recognition
Unsupervised Motion Representation Learning with Capsule Autoencoders
Ziwei Xu, Xudong Shen, Yongkang Wong, Mohan Kankanhalli
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 |
|---|---|---|---|
| snippet length 8 vs 4 (ablation) | MCAE · T20 | accuracy (%): 41.01 -> 69.30 | Improved |
| snippet length 8 vs 16 (ablation) | MCAE · T20 | accuracy (%): 45.83 -> 69.30 | Improved |
| 8 snippet capsules vs 16 (ablation) | MCAE · T20 | accuracy (%): 48.11 -> 69.30 | Improved |
| 8 snippet capsules vs 2 (ablation) | MCAE · T20 | accuracy (%): 64.02 -> 69.30 | Improved |
| 80 segment capsules vs 32 (ablation) | MCAE · T20 | accuracy (%): 42.36 -> 69.30 | Improved |
| 128 segment capsules vs 80 (ablation) | MCAE · T20 | accuracy (%): 69.30 -> 69.44 | No clear effect |
| add smoothness regularizer (ablation) | MCAE · T20 | accuracy (%): 67.60 -> 69.30 | Improved |
| add sparsity regularizer on segment activations (ablation) | MCAE · T20 | accuracy (%): 65.92 -> 69.30 | Improved |
| MCAE capsule model vs LSTM/1D-Conv backbones with far more parameters | MCAE · T20 | accuracy (%): 57.58 -> 69.30 | Improved |
Closest papers by meaning
- MotionBERT: A Unified Perspective on Learning Human Motion Representations (ICCV 2023 · similarity 0.86)
- Action Recognition Using Rate-Invariant Analysis of Skeletal Shape Trajectories (TPAMI 2016 · similarity 0.86)
- Skeleton Motion Words for Unsupervised Skeleton-Based Temporal Action Segmentation (ICCV 2025 · similarity 0.86)
- Global-Local Motion Transformer for Unsupervised Skeleton-Based Action Learning (ECCV 2022 · similarity 0.85)
- PREDICT & CLUSTER: Unsupervised Skeleton Based Action Recognition (CVPR 2020 · similarity 0.85)
- Self-Supervised Skeleton-Based Action Representation Learning: A Benchmark and Beyond (IJCV 2026 · 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.
