Research map / Label-efficient and robust learning / Recognition and long-tail
Trunk-Branch Ensemble Convolutional Neural Networks for Video-Based Face Recognition
Changxing Ding, Dacheng Tao
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 simulated blurred video frames to still-image training data (TS vs SI) (ablation) | TBE-CNN trunk · PaSC handheld set | verification rate at 1% FAR (%): 68.20 -> 79.33 | Improved |
| add simulated blurred video frames to still-image training data (TS vs SI) (ablation) | TBE-CNN trunk · PaSC control set | verification rate at 1% FAR (%): 83.72 -> 87.31 | Improved |
| use simulated blurred video (SV) instead of still images (SI) (ablation) | trunk CNN · PaSC handheld set | verification rate at 1% FAR (%): 68.20 -> 73.80 | Improved |
| use simulated blurred video instead of still images (ablation) | trunk CNN · PaSC control set | verification rate at 1% FAR (%): 83.72 -> 81.89 | Hurt |
| train from scratch on TS data instead of fine-tuning with real video (FT) (ablation) | trunk CNN · PaSC handheld set | verification rate at 1% FAR (%): 73.70 -> 79.33 | Improved |
| add mean-distance regularisation (MDR-TL instead of triplet loss) (ablation) | TBE-CNN · PaSC handheld set (SI data, semi-hard) | verification rate at 1% FAR (%): 87.00 -> 89.60 | Improved |
| add mean-distance regularisation (MDR-TL instead of triplet loss) (ablation) | TBE-CNN · PaSC handheld set (TS data, semi-hard) | verification rate at 1% FAR (%): 92.46 -> 93.91 | Improved |
| add mean-distance regularisation (MDR-TL instead of triplet loss) (ablation) | TBE-CNN · PaSC handheld set (TS data, hardest) | verification rate at 1% FAR (%): 91.47 -> 93.12 | Improved |
| use hardest negative mining instead of semi-hard (ablation) | triplet loss / MDR-TL · PaSC | verification rate at 1% FAR (%): 92.46 -> 91.47 | Hurt |
| use hard instead of semi-hard negative mining (ablation) | triplet loss · PaSC handheld set (TS data) | verification rate at 1% FAR (%): 92.46 -> 92.73 | No clear effect |
| add trunk-branch fusion of facial component patches (TBE-CNN vs trunk network) without BN (ablation) | TBE-CNN · PaSC handheld set | verification rate at 1% FAR (%): 5.82% margin | Improved |
| add trunk-branch fusion of facial component patches with BN layers (ablation) | TBE-CNN · PaSC handheld set | verification rate at 1% FAR (%): 3.48% margin | Improved |
Closest papers by meaning
- Neural Aggregation Network for Video Face Recognition (CVPR 2017 · similarity 0.87)
- Unsupervised Domain Adaptation for Face Recognition in Unlabeled Videos (ICCV 2017 · similarity 0.84)
- Unconstrained Still/Video-Based Face Verification with Deep Convolutional Neural Networks (IJCV 2017 · similarity 0.84)
- Robust RGB-D Face Recognition Using Attribute-Aware Loss (TPAMI 2019 · similarity 0.84)
- Occlusion Robust Face Recognition Based on Mask Learning With Pairwise Differential Siamese Network (ICCV 2019 · similarity 0.84)
- Learning Discriminative Aggregation Network for Video-Based Face Recognition (ICCV 2017 · similarity 0.84)
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.
