Research map / Label-efficient and robust learning / Domain adaptation and generalization
Towards Privacy-Protected Generalized Gaze Estimation Using Diffusion Models and Domain Stability Adaptation Framework
Ziyi Wang, Shengcheng Ye, Faming Fang, Haichuan Song
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 |
|---|---|---|---|
| fine-tune on synthetic gaze data (10000 DDPM+LDM images) (ablation) | PureGaze · Average over 4 cross-dataset tasks | angular error (deg): 8.29 -> 6.93 | Improved |
| increase number of synthetic fine-tuning images (D+L) (ablation) | PureGaze · Average over 4 cross-dataset tasks | angular error (deg): 7.60 (1000) -> 7.33 (5000) -> 6.93 (10000) | Improved |
| mix DDPM-Gaze and LDM-Gaze data instead of one source (10000 images) (ablation) | PureGaze fine-tuning · Average over 4 cross-dataset tasks | angular error (deg): 7.26 (D) / 7.28 (L) -> 6.93 (D+L) | Improved |
| use DSA variance constraint over N=4 same-label images instead of plain fine-tuning (N=1) (ablation) | PureGaze · Average over 4 cross-dataset tasks | angular error (deg): 6.93 -> 6.55 | Improved |
| add DSA to Gaze360 model | Gaze360 UDA model · DG->DD | angular error (deg): 8.77 -> 7.77 | Improved |
| add DSA instead of plain synthetic fine-tuning to UnReGA- | UnReGA- · DG->DM | angular error (deg): 5.58 -> 5.13 (fine-tuning: 5.58) | Improved |
| add plain fine-tuning on synthetic data to Gaze360 model on source-to-source task | Gaze360 UDA model · DG->DM->DG | angular error (deg): 14.43 -> 14.34 | No clear effect |
| use DSA vs best DG baseline GFAL | DSA · ETH-XGaze -> MPIIGaze | angular error (deg): 5.72 -> 6.12 | Hurt |
| use 1000 DDPM-Gaze images only (small amount) for fine-tuning on ETH->EyeDiap (ablation) | PureGaze · ETH-XGaze -> EyeDiap | angular error (deg): 7.48 -> 7.33 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- PureGaze — PureGaze: Purifying Gaze Feature for Generalizable Gaze Estimation (AAAI 2022) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- PureGaze: Purifying Gaze Feature for Generalizable Gaze Estimation (AAAI 2022 · similarity 0.90)
- Learning a Generalized Gaze Estimator from Gaze-Consistent Feature (AAAI 2023 · similarity 0.89)
- Gaze from Origin: Learning for Generalized Gaze Estimation by Embedding the Gaze Frontalization Process (AAAI 2024 · similarity 0.89)
- From Feature to Gaze: A Generalizable Replacement of Linear Layer for Gaze Estimation (CVPR 2024 · similarity 0.86)
- OmniGaze: Reward-inspired Generalizable Gaze Estimation in the Wild (NeurIPS 2025 · similarity 0.86)
- Enhancing 3D Gaze Estimation in the Wild using Weak Supervision with Gaze Following Labels (CVPR 2025 · 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.
