Research map / Video / Action recognition
Multimodal Multipart Learning for Action Recognition in Depth Videos
Amir Shahroudy, Tian-Tsong Ng, Qingxiong Yang, Gang Wang
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 part-based grouping (MP) to plain L2 norm (ablation) | regression-based action classifier · MSR-DailyActivity (subject-wise cross-validation, 252 splits) | accuracy (%): 80.61% -> 81.55% | Improved |
| add modality grouping via hierarchical mixed norm (MMMP) over MP (ablation) | regression-based action classifier · MSR-DailyActivity (subject-wise cross-validation, 252 splits) | accuracy (%): 81.55% -> 84.03% | Improved |
| use part-grouped MP instead of plain L2 (ablation) | regression-based action classifier · MSR-DailyActivity (standard split) | accuracy (%): 87.50% -> 88.13% | Improved |
| use hierarchical mixed norm MMMP instead of MP (ablation) | regression-based action classifier · MSR-DailyActivity (standard split) | accuracy (%): 88.13% -> 91.25% | Improved |
| use plain L2 instead of plain L1 norm (ablation) | regression-based action classifier · MSR-DailyActivity (standard split) | accuracy (%): 86.88% -> 87.50% | Improved |
| use higher-order inner norm (L1/L2/L6) instead of L4 (ablation) | MMMP · MSR-DailyActivity (standard split) | accuracy (%): 91.25% | No clear effect |
| add LOP modality to skeleton features (ablation) | MMMP · MSR-DailyActivity (standard split) | accuracy (%): 79.38% -> 88.13% | Improved |
| add HON4D modality to skeleton+LOP (ablation) | MMMP · MSR-DailyActivity (standard split) | accuracy (%): 88.13% -> 91.25% | Improved |
| use multipart learning on LOP instead of Actionlet Ensemble | MP · MSR-DailyActivity (LOP only) | accuracy (%): 61% -> 79.38% | Improved |
| use MMMP instead of Actionlet Ensemble on skeleton+LOP | MMMP · MSR-DailyActivity (standard split) | accuracy (%): 85.75% -> 88.13% | Improved |
| use MMMP instead of MMTW on skeleton+HON4D | MMMP · MSR-DailyActivity (standard split) | accuracy (%): 88.75% -> 89.38% | Improved |
| use MMMP instead of best prior on MSR-Action3D (protocol 1) | MMMP · MSR-Action3D (three subsets) | accuracy (%): 96.7% -> 98.2% | Improved |
Closest papers by meaning
- Deep Multimodal Feature Analysis for Action Recognition in RGB+D Videos (TPAMI 2017 · similarity 0.89)
- MMNet: A Model-Based Multimodal Network for Human Action Recognition in RGB-D Videos (TPAMI 2022 · similarity 0.89)
- ASPnet: Action Segmentation With Shared-Private Representation of Multiple Data Sources (CVPR 2023 · similarity 0.87)
- Stitch, Contrast, and Segment: Learning a Human Action Segmentation Model Using Trimmed Skeleton Videos (AAAI 2025 · similarity 0.86)
- DVANet: Disentangling View and Action Features for Multi-View Action Recognition (AAAI 2024 · similarity 0.86)
- Recovering Complete Actions for Cross-dataset Skeleton Action Recognition (NeurIPS 2024 · similarity 0.86)
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.
