Research map / Architectures and efficiency / Mixture of experts
MoEGaze: A Mixture of Experts Approach for Generalizable Gaze Estimation
Zheng Liu, Feng Lu
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 balance loss (ablation) | MoEGaze · DG->DD | angular error: 8.24 -> 7.66 | Improved |
| add balance loss (ablation) | MoEGaze · DG->DM | angular error: 7.17 -> 7.54 | Hurt |
| add entropy regularization (ablation) | MoEGaze · cross-domain average | angular error: 7.15 -> 6.95 | Improved |
| add alignment loss (ablation) | MoEGaze · cross-domain average | angular error: 6.95 -> 6.71 | Improved |
| use N=4,K=3 experts instead of N=4,K=1 (ablation) | MoEGaze · DE->DM and DE->DD average | angular error: 7.92 -> 6.36 | Improved |
| increase experts from 4 to 8 (K=4) (ablation) | MoEGaze · DE->DM and DE->DD average | angular error: 6.36 -> 7.05 | Hurt |
| increase K from 3 to 4 at N=4 (ablation) | MoEGaze · DE->DM and DE->DD average | angular error: 6.36 -> 6.56 | No clear effect |
| vary aggregation layers L (1,4,6,12) (ablation) | MoEGaze · DE->DM average | angular error: 7.17, 7.01, 6.64, 7.37 | Mixed |
| vary attention heads H (1,4,8,16) (ablation) | MoEGaze · DE->DM and DE->DD average | angular error: 6.88, 6.62, 6.36, 6.56 | Mixed |
| fine-tune on in-group appearance data vs full data (ablation) | MoEGaze · Gaze (ETH-XGaze, glasses test) | angular error: 7.64 -> 7.19 | Improved |
| train MoEGaze with 10% data vs baseline on full data (ablation) | MoEGaze · DE->DD | angular error | Improved |
Closest papers by meaning
- PureGaze: Purifying Gaze Feature for Generalizable Gaze Estimation (AAAI 2022 · similarity 0.89)
- Gaze from Origin: Learning for Generalized Gaze Estimation by Embedding the Gaze Frontalization Process (AAAI 2024 · similarity 0.88)
- OmniGaze: Reward-inspired Generalizable Gaze Estimation in the Wild (NeurIPS 2025 · similarity 0.87)
- From Feature to Gaze: A Generalizable Replacement of Linear Layer for Gaze Estimation (CVPR 2024 · similarity 0.87)
- CLIP-Gaze: Towards General Gaze Estimation via Visual-Linguistic Model (AAAI 2024 · similarity 0.87)
- 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.
