Research map / 3D vision / Human pose estimation
Unsupervised Representation Learning for Gaze Estimation
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 |
|---|---|---|---|
| Pretrain GResNetHP convolutional layers with unsupervised gaze representation learning | GResNetHP · Eyediap | gaze error (degrees): 0.2^{\circ}\sim0.3^{\circ} | Helped |
| Pretrain GResNetHP convolutional layers with unsupervised gaze representation learning | GResNetHP · Columbia Gaze | gaze error (degrees): 0.2^{\circ}\sim0.3^{\circ} | Helped |
| Pretrain GResNetHP convolutional layers with unsupervised gaze representation learning | GResNetHP · UTMultiview | gaze error (degrees): 0.2^{\circ}\sim0.3^{\circ} | Helped |
| Remove random scaling and translation augmentation (ablation) | Unsupervised gaze representation learning · Few-shot gaze estimation | gaze error (degrees): \sim \!\! 3^{\circ} | Hurt |
| Add random scaling and translation augmentation (ablation) | Unsupervised gaze representation learning · Few-shot gaze estimation | gaze error (degrees): \sim \!\! 3^{\circ} | Helped |
| Remove warping field regularization (ablation) | Unsupervised gaze representation learning · Few-shot gaze estimation | gaze error (degrees) | Hurt |
| Add warping field regularization (ablation) | Unsupervised gaze representation learning · Few-shot gaze estimation | gaze error (degrees) | Helped |
| Replace GResNet with the shallow MnistNet gaze network (ablation) | Unsupervised gaze estimation · Few-shot gaze estimation | gaze error (degrees) | Hurt |
| Use unsupervised gaze representation learning with no more than 100 calibration samples | Few-shot gaze estimation · Eyediap, Columbia Gaze, and UTMultiview | gaze error (degrees): $7^{\circ}\sim8^{\circ}$ | Helped |
| Use unsupervised representation learning followed by adaptation rather than direct training on calibration samples | Few-shot gaze estimation · Eyediap, Columbia Gaze, and UTMultiview | gaze error (degrees) | Helped |
| Use unsupervised SVR adaptation instead of the supervised-UTMultiview SVR adaptation | Cross-dataset gaze estimation · UTMultiview to Columbia Gaze | gaze error (degrees) | Helped |
| Increase calibration samples for linear adaptation rather than network fine-tuning | Few-shot gaze estimation · Few-shot gaze estimation | gaze error (degrees) | Mixed |
Closest papers by meaning
- Enhancing 3D Gaze Estimation in the Wild using Weak Supervision with Gaze Following Labels (CVPR 2025 · similarity 0.90)
- De^2Gaze: Deformable and Decoupled Representation Learning for 3D Gaze Estimation (CVPR 2025 · similarity 0.89)
- Unsupervised Gaze Representation Learning from Multi-view Face Images (CVPR 2024 · similarity 0.88)
- Gaze from Origin: Learning for Generalized Gaze Estimation by Embedding the Gaze Frontalization Process (AAAI 2024 · similarity 0.88)
- GA3CE: Unconstrained 3D Gaze Estimation with Gaze-Aware 3D Context Encoding (CVPR 2025 · similarity 0.88)
- 3D Prior Is All You Need: Cross-Task Few-shot 2D Gaze Estimation (CVPR 2025 · similarity 0.87)
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.