Research map / Image generation / GANs
On Diverse Asynchronous Activity Anticipation
He Zhao, Richard P. Wildes
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 normalized distance regularizer to adversarial training (alpha=1) (ablation) | Gumbel GAN, LSTM · Breakfast P2 | accuracy (%): 60.98 -> 68.96 | Improved |
| add normalized distance regularizer to adversarial training (alpha=1) (ablation) | Gumbel GAN, LSTM · Breakfast P2 | LL: -6.77 -> -4.83 | Improved |
| add normalized distance regularizer to adversarial training (alpha=1) (ablation) | Gumbel GAN, LSTM · Breakfast P2 | diversity (cosine distance): 0.124 -> 0.262 | Improved |
| add normalized distance regularizer (alpha=0.1) (ablation) | Gumbel GAN, LSTM · Breakfast P2 | accuracy (%): 23.01 -> 27.61 | Improved |
| add normalized distance regularizer (alpha=10) (ablation) | Gumbel GAN, LSTM · Breakfast P2 | accuracy (%): 35.19 -> 36.64 | Improved |
| vary Gumbel temperature alpha without regularizer (ablation) | Gumbel GAN, LSTM · Breakfast P2 | accuracy (%): 23.01 / 60.98 / 35.19 / 33.78 / 19.97 | Mixed |
| increase number of samples from 2 to 16 (ablation) | Ours-LSTM (Avg) · Breakfast P1 (obs 20%, pred 10%) | MoC accuracy: 0.6157 -> 0.7222 | Improved |
| increase number of samples from 16 to 64 (ablation) | Ours-LSTM (Avg) · Breakfast P1 (obs 20%, pred 10%) | MoC accuracy: 0.7222 -> 0.7572 | No clear effect |
| increase adversarial training epochs beyond 1500 (ablation) | Ours-LSTM · Epic-Kitchens | accuracy at long prediction times | Hurt |
| add adversarial training after MLE pretraining (ablation) | Ours-LSTM · Epic-Kitchens | accuracy | Improved |
| use max over 16 samples instead of average | Ours-LSTM · Breakfast P1 (obs 20%, pred 10%) | MoC accuracy: 0.7222 -> 0.8208 | Improved |
| use Ours-LSTM instead of TOS-Dense | Ours-LSTM (Avg) · Breakfast P1 (obs 20%, pred 10%) | MoC accuracy: 0.6446 -> 0.7222 | Improved |
Closest papers by meaning
- Predicting the Future: A Jointly Learnt Model for Action Anticipation (ICCV 2019 · similarity 0.87)
- A Hybrid Egocentric Activity Anticipation Framework via Memory-Augmented Recurrent and One-Shot Representation Forecasting (CVPR 2022 · similarity 0.86)
- Rethinking Learning Approaches for Long-Term Action Anticipation (ECCV 2022 · similarity 0.85)
- Adversarial Generative Grammars for Human Activity Prediction (ECCV 2020 · similarity 0.85)
- Time-Conditioned Action Anticipation in One Shot (CVPR 2019 · similarity 0.85)
- FutureHuman3D: Forecasting Complex Long-Term 3D Human Behavior from Video Observations (CVPR 2024 · 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.
