Research map / 3D vision / Human pose estimation
Constrained Rotation Optimization: Revisiting Crop-Based Gaze Estimation
Riccardo Santambrogio, Jiawei Qin, Matteo Matteucci, Yusuke Sugano
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 |
|---|---|---|---|
| replace normalization with CROp crops | gaze ResNet18 · MPII->XG cross-dataset (Det) | angular error (deg): 30.55 -> 29.56 | Improved |
| replace normalization with CROp crops | gaze ResNet18 · EVE->ED cross-dataset (GT) | angular error (deg): 21.32 -> 18.84 | Improved |
| replace normalization with CROp crops | gaze ResNet18 · XG->MPII cross-dataset (GT) | angular error (deg): 7.19 -> 7.41 | Hurt |
| replace normalization with CROp crops | gaze ResNet18 · MPII within-dataset (GT) | angular error (deg): 4.93 -> 5.02 | Hurt |
| replace normalization with CROp crops | gaze ResNet18 · XG within-dataset (Det) | angular error (deg): 6.45 -> 5.55 | Improved |
| add head pose supervision to CROp (ablation) | CROp gaze model · MPII->XG cross-dataset (Det) | angular error (deg): 28.29 -> 29.56 | Hurt |
| add head pose supervision to CROp (ablation) | CROp gaze model · EVE->ED cross-dataset (Det) | angular error (deg): 21.13 -> 18.47 | Improved |
| add head pose supervision to CROp (ablation) | CROp gaze model · XG->MPII cross-dataset (Det) | angular error (deg): 8.05 -> 7.76 | Improved |
| add head pose supervision to normalization (ablation) | Norm gaze model · XG->ED cross-dataset (Det) | angular error (deg): 13.19 -> 10.55 | Improved |
| use CROp instead of normalization at 40% bbox noise | gaze ResNet18 · EVE->MPII bbox noise | angular error (deg): 9.56 -> 10.25 | Hurt |
| use CROp instead of normalization at 50% occlusion | gaze ResNet18 · EVE->MPII occlusion | angular error (deg): 11.17 -> 9.90 | Improved |
| replace normalization pipeline with CROp pipeline | gaze pipeline · XGaze runtime | total time (ms): 20.74 -> 15.97 | Improved |
Closest papers by meaning
- Multi-scale Object-Aware Gaze Estimation via Geometric Reasoning (ECCV 2026 · similarity 0.86)
- RT-GENE: Real-Time Eye Gaze Estimation in Natural Environments (ECCV 2018 · similarity 0.86)
- MPIIGaze: Real-World Dataset and Deep Appearance-Based Gaze Estimation (TPAMI 2017 · similarity 0.86)
- 3DGazeNet: Generalizing Gaze Estimation with Weak Supervision from Synthetic Views (ECCV 2024 · similarity 0.85)
- From Feature to Gaze: A Generalizable Replacement of Linear Layer for Gaze Estimation (CVPR 2024 · similarity 0.85)
- Gaze from Origin: Learning for Generalized Gaze Estimation by Embedding the Gaze Frontalization Process (AAAI 2024 · similarity 0.83)
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.
