Research map / Architectures and efficiency / CNN design
Incremental Boosting Convolutional Neural Network for Facial Action Unit Recognition
Shizhong Han, Zibo Meng, AHMED-SHEHAB KHAN, Yan Tong
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 |
|---|---|---|---|
| replace CNN with IB-CNN | IB-CNN · SEMAINE | average F1: 0.347 -> 0.416 | Improved |
| replace CNN with IB-CNN | IB-CNN · SEMAINE | average 2AFC: 0.735 -> 0.775 | Improved |
| replace boosting CNN (B-CNN) with IB-CNN, i.e. add incremental accumulation over time | IB-CNN · SEMAINE | average F1: 0.310 -> 0.416 | Improved |
| use only strong-classifier loss (IB-CNN-S) instead of strong plus weak losses (ablation) | IB-CNN · SEMAINE | average F1: 0.412 -> 0.416 | Improved |
| IB-CNN on the rarest AU28 versus CNN and B-CNN | IB-CNN · SEMAINE AU28 | F1: 0.280 -> 0.490 | Improved |
| plain B-CNN versus CNN on infrequent AUs | B-CNN · SEMAINE | average F1: 0.347 -> 0.310 | Hurt |
| replace CNN with IB-CNN | IB-CNN · DISFA | average F1: 0.405 -> 0.457 | Improved |
| replace B-CNN with IB-CNN | IB-CNN · DISFA | average 2AFC: 0.815 -> 0.823 | Improved |
| IB-CNN on infrequent AU5 versus CNN | IB-CNN · DISFA AU5 | F1: 0.195 -> 0.312 | Improved |
| IB-CNN on AU15 versus CNN | IB-CNN · DISFA AU15 | F1: 0.174 -> 0.135 | Hurt |
| vary eta scale c from 0.5 to 16 (ablation) | IB-CNN · SEMAINE | average F1 | No clear effect |
| vary number of IB-layer input neurons 64 to 2048 (ablation) | IB-CNN · SEMAINE | F1 | No clear effect |
Closest papers by meaning
- Deep Facial Action Unit Recognition From Partially Labeled Data (ICCV 2017 · similarity 0.89)
- Deep Structure Inference Network for Facial Action Unit Recognition (ECCV 2018 · similarity 0.88)
- Weakly-Supervised Deep Convolutional Neural Network Learning for Facial Action Unit Intensity Estimation (CVPR 2018 · similarity 0.88)
- Optimizing Filter Size in Convolutional Neural Networks for Facial Action Unit Recognition (CVPR 2018 · similarity 0.88)
- Facial Action Unit Detection With Transformers (CVPR 2021 · similarity 0.87)
- Action Unit Detection With Region Adaptation, Multi-Labeling Learning and Optimal Temporal Fusing (CVPR 2017 · 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.
