Research map / 3D vision / Human pose estimation
Few-Shot Adaptive Gaze Estimation
Seonwook Park, Shalini De Mello, Pavlo Molchanov, Umar Iqbal, Otmar Hilliges, Jan Kautz
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 |
|---|---|---|---|
| use MAML instead of naive fine-tuning for the person-specific gaze estimator (ablation) | AE features gaze estimator · MPIIGaze (train on GazeCapture) | gaze error (degrees): 7.17 -> 6.61 (k=1) | Improved |
| replace autoencoder with rotation-aware disentangling DT-ED (ablation) | MAML gaze estimator · MPIIGaze (k=9) | gaze error (degrees): 5.62 -> 4.87 | Improved |
| add gaze supervision loss to reconstruction loss (ablation) | DT-ED · MPIIGaze (k=9) | gaze error (degrees): 4.87 -> 3.60 | Improved |
| add person-specific embedding consistency loss to reconstruction loss (ablation) | DT-ED · MPIIGaze (k=9) | gaze error (degrees): 4.87 -> 3.32 | Improved |
| combine reconstruction, gaze and embedding consistency losses (ablation) | DT-ED · MPIIGaze (k=9) | gaze error (degrees): 4.87 -> 3.14 | Improved |
| use person-independent embedding consistency (also minimize inter-person differences) instead of person-specific (ablation) | DT-ED · MPIIGaze | gaze error (degrees) | Hurt |
| add triplet loss maximizing inter-person feature differences (ablation) | DT-ED embedding consistency · MPIIGaze | gaze error (degrees) | Hurt |
| use FAZE (DT-ED + MAML) instead of DenseNet/VGG-16 CNN trained on gaze objective + MAML | MAML gaze estimator · MPIIGaze | gaze error (degrees) | Improved |
| use FAZE instead of Liu et al. differential gaze re-implementation | gaze personalization · MPIIGaze (k=9) | gaze error (degrees): 3.53 -> 3.14 | Improved |
| use FAZE instead of polynomial fit to PoR | gaze personalization · MPIIGaze (k=9) | gaze error (degrees): 4.19 -> 3.14 | Improved |
| use FAZE in within-MPIIGaze leave-one-person-out vs Liu et al. | gaze personalization · MPIIGaze (within-dataset, k=9) | gaze error (degrees): 4.67 -> 3.88 | Improved |
| set latent gaze code dimension F_g=2 rather than 16 or 3 (ablation) | FAZE · GazeCapture test and MPIIGaze | gaze error (degrees) | Improved |
Closest papers by meaning
- Test-Time Personalization with Meta Prompt for Gaze Estimation (AAAI 2024 · similarity 0.88)
- Gaze-LLE: Gaze Target Estimation via Large-Scale Learned Encoders (CVPR 2025 · similarity 0.88)
- Sharingan: A Transformer Architecture for Multi-Person Gaze Following (CVPR 2024 · similarity 0.87)
- Gaze from Origin: Learning for Generalized Gaze Estimation by Embedding the Gaze Frontalization Process (AAAI 2024 · similarity 0.87)
- Enhancing 3D Gaze Estimation in the Wild using Weak Supervision with Gaze Following Labels (CVPR 2025 · similarity 0.87)
- CLIP-Gaze: Towards General Gaze Estimation via Visual-Linguistic Model (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.
