Research map / Image generation / Autoregressive and latent generation
Self-Learning Transformations for Improving Gaze and Head Redirection
Read the paperCheck 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 |
|---|---|---|---|
| self-predicted pseudo labels for transformation (ablation) | T-ED base model · GazeCapture | gaze redirection error: 7.114 -> 5.107 | Helped |
| pseudo labels for transformation (ablation) | T-ED base model · GazeCapture | g->h disentanglement error: 0.542 -> 0.660 | Hurt |
| discover extraneous factors f_u (ablation) | ST-ED · GazeCapture | gaze redirection error: 5.107 -> 4.716 | Helped |
| add functional loss and disentanglement loss (ablation) | ST-ED + f_u · GazeCapture | gaze redirection error: 4.716 -> 2.195 | Helped |
| add functional and disentanglement losses (ablation) | ST-ED + f_u · GazeCapture | g->h disentanglement error: 0.385 -> 0.388 | No clear effect |
| align all extraneous factors to target pseudo labels at eval (ablation) | ST-ED · GazeCapture | LPIPS: 0.248 -> 0.205 | Helped |
| ST-ED vs StarGAN | ST-ED · GazeCapture | gaze redirection error: 4.602 -> 2.195 | Helped |
| ST-ED vs He et al. on Columbia LPIPS | ST-ED · Columbia | LPIPS: 0.227 -> 0.242 | Hurt |
| halve loss weights for L_R, L_F or L_PL (ablation) | ST-ED · GazeCapture | gaze redirection error: 2.195 -> 2.231/2.321/2.249 | No clear effect |
| remove embedding losses (ablation) | ST-ED · GazeCapture | gaze redirection error: 2.195 -> 2.359 | Mixed |
| augment 2.5k labeled data with ST-ED redirected images | gaze estimator · cross-dataset gaze estimation | gaze error: 6.138 -> 5.203 (GazeCapture) | Helped |
| augment with He et al. redirected images | gaze estimator · cross-dataset gaze estimation | gaze error: 6.138 -> 9.882 (GazeCapture); 10.536 -> 9.274 (Columbia) | Mixed |
Closest papers by meaning
- Bootstrap AutoEncoders With Contrastive Paradigm for Self-supervised Gaze Estimation (ICML 2024 · similarity 0.88)
- OmniGaze: Reward-inspired Generalizable Gaze Estimation in the Wild (NeurIPS 2025 · similarity 0.87)
- ST-DiffEye: Diffusion-based Continuous Gaze Generation via Joint Scanpath-Trajectory Modeling (NeurIPS 2026 · similarity 0.86)
- Enhancing 3D Gaze Estimation in the Wild using Weak Supervision with Gaze Following Labels (CVPR 2025 · similarity 0.86)
- Gaze from Origin: Learning for Generalized Gaze Estimation by Embedding the Gaze Frontalization Process (AAAI 2024 · similarity 0.86)
- DiffEye: Diffusion-Based Continuous Eye-Tracking Data Generation Conditioned on Natural Images (NeurIPS 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.