Research map / Video / Action recognition
Evidential Deep Learning for Open Set Action Recognition
Wentao Bao, Qi Yu, Yu Kong
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 EUC calibration loss (ablation) | DEAR (TPN) · Ablation, UCF-101 + HMDB-51 (unknown) | Open maF1 (%): 74.95 -> 75.88 | Improved |
| add CED debiasing module (joint training) (ablation) | DEAR (TPN) · Ablation, UCF-101 + HMDB-51 (unknown) | Open maF1 (%): 75.88 -> 81.79 | Improved |
| use joint rather than alternative training of CED (ablation) | DEAR (TPN) · Ablation, UCF-101 + HMDB-51 (unknown) | Open maF1 (%): 81.18 -> 81.79 | Improved |
| add EUC loss (ablation) | DEAR · Open Set (K+1) calibration | ECE: 0.284 -> 0.268 | Improved |
| add EUC loss (ablation) | DEAR · Closed Set (K) calibration | ECE: 0.030 -> 0.029 | Improved |
| add CED module (ablation) | DEAR (TSM) · Mimetics (unbiased) | top-1 accuracy (%): 26.56 -> 34.38 | Improved |
| add CED module (ablation) | DEAR (TSM) · Kinetics (biased) | top-1 accuracy (%): 91.18 -> 91.18 | No clear effect |
| replace SoftMax with DEAR | I3D · UCF-101 + HMDB-51 (unknown) | Open maF1 (%): 73.19 -> 77.24 | Improved |
| replace SoftMax with DEAR | SlowFast · UCF-101 + HMDB-51 (unknown) | Open maF1 (%): 78.04 -> 85.48 | Improved |
| replace OpenMax with DEAR | TSM · UCF-101 + MiT-v2 (unknown) | Open maF1 (%): 71.81 -> 70.15 | Hurt |
| replace SoftMax with DEAR | TPN · UCF-101 + MiT-v2 (unknown) | Open Set AUC (%): 81.35 -> 81.80 | Improved |
Closest papers by meaning
- Open Set Action Recognition via Multi-Label Evidential Learning (CVPR 2023 · similarity 0.90)
- Bayesian Evidential Deep Learning for Online Action Detection (ECCV 2024 · similarity 0.89)
- Uncertainty-Aware Dual-Evidential Learning for Weakly-Supervised Temporal Action Localization (TPAMI 2023 · similarity 0.88)
- Bayesian Uncertainty Attribution-Guided Fine-Tuning for Open-Set Action Recognition (ECCV 2026 · similarity 0.87)
- OpenTAL: Towards Open Set Temporal Action Localization (CVPR 2022 · similarity 0.87)
- Towards Evidential and Class Separable Open Set Object Detection (AAAI 2024 · similarity 0.86)
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.
