Research map / Video / Action recognition
Generating Human Action Videos by Coupling 3D Game Engines and Probabilistic Graphical Models
César Roberto de Souza, Adrien Gaidon, Yohann Cabon, Naila Murray, Antonio Manuel López
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 PHAV multi-task training (Cool-TSN) to TSN | TSN · UCF-101 RGB+Flow | mean accuracy (%): 93.6 -> 94.2 | Improved |
| add PHAV multi-task training (Cool-TSN) to TSN | TSN · HMDB-51 RGB+Flow | mean accuracy (%): 66.6 -> 69.5 | Improved |
| add PHAV pretraining and fine-tuning (TSN-FT) to TSN | TSN · HMDB-51 RGB+Flow | mean accuracy (%): 66.6 -> 68.9 | Improved |
| add PHAV pretraining and fine-tuning (TSN-FT) to TSN | TSN · UCF-101 RGB+Flow | mean accuracy (%): 93.6 -> 94.1 | Improved |
| use multi-task Cool-TSN instead of PHAV fine-tuning | TSN-FT · HMDB-51 RGB (spatial) | mean accuracy (%): 51.0 -> 53.0 | Improved |
| add PHAV videos with 1% of real HMDB-51 training data | TSN · HMDB-51 split 1, 1% real data | mean accuracy (%): 8.1 -> 12.7 | Improved |
| add PHAV videos with 25% of real HMDB-51 training data | TSN · HMDB-51 split 1, 25% real data | mean accuracy (%): 54.8 -> 60.7 | Improved |
| add PHAV videos with 50% of real HMDB-51 data | TSN · HMDB-51 split 1, 50% real data | mean accuracy (%): 62.9 -> 65.8 | Improved |
| add PHAV videos with 10% of real UCF-101 data | TSN · UCF-101 split 1, 10% real data | mean accuracy (%): 80.9 -> 84.4 | Improved |
| add PHAV videos with 50% of real UCF-101 data | TSN · UCF-101 split 1, 50% real data | mean accuracy (%): 92.5 -> 92.7 | Improved |
| add PHAV videos with full real UCF-101 data | TSN · UCF-101 split 1, 100% real data | mean accuracy (%): 92.8 -> 93.3 | Improved |
| double procedural noise (failure cases) in PHAV from 10% to 20% (ablation) | Cool-TSN · UCF-101 RGB+Flow | mean accuracy (%): 94.2 -> 94.2 | No clear effect |
Closest papers by meaning
- Procedural Generation of Videos to Train Deep Action Recognition Networks (CVPR 2017 · similarity 0.97)
- Synthetic Humans for Action Recognition from Unseen Viewpoints (IJCV 2021 · similarity 0.91)
- Generative Action Tell-Tales: Assessing Human Motion in Synthesized Videos (ECCV 2026 · similarity 0.90)
- Learning Video Representations of Human Motion From Synthetic Data (CVPR 2022 · similarity 0.89)
- HumanVid: Demystifying Training Data for Camera-controllable Human Image Animation (NeurIPS 2024 · similarity 0.87)
- DiT4DiT: Jointly Modeling Video Dynamics and Actions for Generalizable Robot Control (NeurIPS 2026 · 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.
