Research map / Label-efficient and robust learning / Domain adaptation and generalization
Generalizing Hand Segmentation in Egocentric Videos With Uncertainty-Guided Model Adaptation
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use uncertainty-guided self-training with Bayesian CNN estimates (ablation) | Hand segmentation model trained on EGTEA · GTEA, EDSH-2, EDSH-K, UTG, YHG, Egohands | mIoU: 0.8896 -> 0.8945; 0.7632 -> 0.7965; 0.7576 -> 0.7812; 0.5832 -> 0.6762; 0.3619 -> 0.5223; 0.4235 -> 0.4665 | Helped |
| Add hand-shape adversarial constraint (ablation) | Bayesian CNN+uma · GTEA, EDSH-2, EDSH-K, UTG | mIoU: 0.8945 -> 0.8990; 0.7965 -> 0.8025; 0.7812 -> 0.7951; 0.6762 -> 0.6827 | Helped |
| Add hand-shape adversarial constraint (ablation) | Bayesian CNN+uma · YHG | mIoU: 0.5223 -> 0.5596 | Helped |
| Add hand-shape adversarial constraint (ablation) | Bayesian CNN+uma · Egohands | mIoU: 0.4665 -> 0.4660 | Hurt |
| Use Bayesian CNN instead of standard CNN (ablation) | Hand segmentation model trained on EGTEA · GTEA, EDSH-2, EDSH-K, UTG, YHG | mIoU: 0.8845 -> 0.8896; 0.6936 -> 0.7632; 0.7205 -> 0.7576; 0.5481 -> 0.5832; 0.2831 -> 0.3619 | Helped |
| Use Bayesian CNN instead of standard CNN (ablation) | Hand segmentation model trained on EGTEA · Egohands | mIoU: 0.4019 -> 0.4235 | Helped |
| Apply uncertainty-guided adaptation with standard-CNN uncertainty (ablation) | Standard CNN hand segmentation · GTEA | mIoU: 0.8845 -> 0.8766 | Hurt |
| Apply uncertainty-guided adaptation with standard-CNN uncertainty (ablation) | Standard CNN hand segmentation · EDSH-2, EDSH-K, UTG, YHG, Egohands | mIoU: 0.6936 -> 0.7141; 0.7205 -> 0.7723; 0.5481 -> 0.6089; 0.2831 -> 0.3159; 0.4019 -> 0.4252 | Helped |
| Use proposed uncertainty-guided adaptation and hand-shape prior | Hand segmentation model trained on EGTEA · GTEA, EDSH-2, EDSH-K, UTG, YHG, Egohands | IoU: 0.8845 -> 0.8990; 0.6936 -> 0.8025; 0.7205 -> 0.7951; 0.5481 -> 0.6827; 0.2831 -> 0.5596; 0.4019 -> 0.4660 | Helped |
| Continue iterative adaptation beyond its best-performing point (ablation) | Bayesian CNN+uma and Bayesian CNN+uma+hs · Cross-dataset hand segmentation | IoU | Hurt |
Closest papers by meaning
- Analysis of Hand Segmentation in the Wild (CVPR 2018 · similarity 0.91)
- Modeling Fine-Grained Hand-Object Dynamics for Egocentric Video Representation Learning (ICLR 2025 · similarity 0.87)
- Cascaded Interactional Targeting Network for Egocentric Video Analysis (CVPR 2016 · similarity 0.87)
- H+O: Unified Egocentric Recognition of 3D Hand-Object Poses and Interactions (CVPR 2019 · similarity 0.87)
- Single-to-Dual-View Adaptation for Egocentric 3D Hand Pose Estimation (CVPR 2024 · similarity 0.86)
- Estimating Ego-Body Pose from Doubly Sparse Egocentric Video Data (NeurIPS 2024 · 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.