Research map / Video / Action recognition
Deep Multimodal Feature Analysis for Action Recognition in RGB+D Videos
Amir Shahroudy, Tian-Tsong Ng, Yihong Gong, 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 modality-specific components Z to the shared network (ablation) | DSSCA SSLM local, 1 layer · Online RGBD Action S3 | accuracy (%): 73.0% -> 76.6% | Improved |
| add modality-specific components Z to the shared network (ablation) | DSSCA SSLM local, 2 layers · Online RGBD Action S3 | accuracy (%): 73.9% -> 77.5% | Improved |
| stack local and holistic networks instead of holistic only (ablation) | DSSCA SSLM · NTU RGB+D (cross-subject) | accuracy (%): 70.4% -> 74.9% | Improved |
| stack local and holistic networks instead of local only (ablation) | DSSCA SSLM · NTU RGB+D (cross-subject) | accuracy (%): 66.4% -> 74.9% | Improved |
| stack local and holistic networks instead of holistic only (ablation) | DSSCA SSLM · Online RGBD Action S1 | accuracy (%): 92.0% -> 95.5% | Improved |
| stack local and holistic networks instead of local only (ablation) | DSSCA SSLM · RGBD-HuDaAct | accuracy (%): 98.7% -> 99.0% | No clear effect |
| use SSLM classifier instead of kernel combination on DSSCA components | DSSCA · Online RGBD Action S1 | accuracy (%): 92.9% -> 95.5% | Improved |
| use SSLM classifier instead of kernel combination on DSSCA components | DSSCA · 3D Action Pairs | accuracy (%): 100.0% -> 100.0% | No clear effect |
| replace kernel combination of raw features with DSSCA factorized components | multi-kernel SVM · Online RGBD Action S3 | accuracy (%): 80.2% -> 82.0% | Improved |
| replace descriptor concatenation with DSSCA SSLM | linear SVM · NTU RGB+D (cross-subject) | accuracy (%): 59.7% -> 74.9% | Improved |
| add modality-specific RICA components to CCA shared components | CCA at atomic level · RGBD-HuDaAct | accuracy (%): 93.9% -> 96.4% | Improved |
| replace CCA-RICA with single-unit SSCA | atomic level analysis · RGBD-HuDaAct | accuracy (%): 96.4% -> 97.9% | Improved |
Closest papers by meaning
- MMNet: A Model-Based Multimodal Network for Human Action Recognition in RGB-D Videos (TPAMI 2022 · similarity 0.90)
- Modality Distillation with Multiple Stream Networks for Action Recognition (ECCV 2018 · similarity 0.89)
- Multimodal Multipart Learning for Action Recognition in Depth Videos (TPAMI 2016 · similarity 0.89)
- Graph Distillation for Action Detection with Privileged Modalities (ECCV 2018 · similarity 0.88)
- Cooperative Training of Deep Aggregation Networks for RGB-D Action Recognition (AAAI 2018 · similarity 0.88)
- A Multimodal, Multi-Task Adapting Framework for Video Action Recognition (AAAI 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.
