Research map / Video / Action recognition
PM-GANs: Discriminative Representation Learning for Action Recognition Using Partial-modalities
Lan Wang, Chenqiang Gao, Luyu Yang, Yue Zhao, Wangmeng Zuo, Deyu Meng
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 generated RGB features to infrared features (ablation) | PM-GANs · Infrared-Visible action dataset (random split) | accuracy: 71.67 -> 78 | Improved |
| use generated RGB representation instead of infrared features alone | PM-GANs · Infrared-Visible action dataset (random split) | accuracy: 71.67 -> 76.67 | Improved |
| use real RGB with infrared instead of generated RGB with infrared | PM-GANs · Infrared-Visible action dataset (random split) | accuracy: 78 -> 82.33 | Improved |
| add generated RGB features to infrared features (ablation) | PM-GANs · Infrared-Visible action dataset (winter test split) | accuracy: 74.17 -> 80.28 | Improved |
| use optical flow input instead of original clips | 3D-CNN on infrared · Infrared-Visible action dataset (3D-CNN features + SVM) | accuracy: 55% -> 69.67% | Improved |
| use optical flow input instead of original clips | 3D-CNN on RGB · Infrared-Visible action dataset (3D-CNN features + SVM) | accuracy: 49% -> 78.66% | Improved |
| fuse RGB with infrared using optical flow features | 3D-CNN · Infrared-Visible action dataset (3D-CNN features + SVM) | accuracy: 69.67% -> 80.67% | Improved |
| use MHI input instead of original clips | 3D-CNN on infrared · Infrared-Visible action dataset (3D-CNN features + SVM) | accuracy: 55% -> 61% | Improved |
| fuse RGB with infrared using original clips | 3D-CNN · Infrared-Visible action dataset (3D-CNN features + SVM) | accuracy: 55% -> 55.33% | Mixed |
| use PM-GANs instead of Two-Stream 3D-CNN | infrared action recognition · Infrared-Visible action dataset | accuracy: 74.67 -> 78 | Improved |
| use 3D-CNN instead of 2D two-stream CNN | action recognition · Infrared-Visible action dataset | accuracy: 68 -> 69.67 | Improved |
Closest papers by meaning
- Adversarial Action Prediction Networks (TPAMI 2018 · similarity 0.87)
- Generative Multi-View Human Action Recognition (ICCV 2019 · similarity 0.86)
- MMNet: A Model-Based Multimodal Network for Human Action Recognition in RGB-D Videos (TPAMI 2022 · similarity 0.86)
- Learning with Privileged Information via Adversarial Discriminative Modality Distillation (TPAMI 2019 · similarity 0.85)
- DVANet: Disentangling View and Action Features for Multi-View Action Recognition (AAAI 2024 · similarity 0.85)
- Prompt-guided Disentangled Representation for Action Recognition (NeurIPS 2025 · similarity 0.83)
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.
