Research map / 3D vision / Human pose estimation
RGBD Based Gaze Estimation via Multi-Task CNN
Dongze Lian, Ziheng Zhang, Weixin Luo, Lina Hu, Minye Wu, Zechao Li, Jingyi Yu, Shenghua Gao
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 head pose features (RGB+depth face branch) (ablation) | Our method · ShanghaiTechGaze+ (RGBD) test | gaze point error (mm): 54.0 -> 38.7 | Improved |
| add head pose features (ablation) | Our method · EYEDIAP | gaze direction error (deg): 7.8 -> 4.8 | Improved |
| add depth input to head pose branch (ablation) | Our method · ShanghaiTechGaze+ (RGBD) test | gaze point error (mm): 46.7 -> 38.7 | Improved |
| add RGB face input to head pose branch (ablation) | Our method · ShanghaiTechGaze+ (RGBD) test | gaze point error (mm): 52.9 -> 38.7 | Improved |
| add GAN-based depth reconstruction decoder (ablation) | Our method · ShanghaiTechGaze+ (RGBD) test | gaze point error (mm): 44.2 -> 38.7 | Improved |
| process RGB and depth in separate CNN branches instead of stacking (ablation) | Our method · ShanghaiTechGaze+ (RGBD) test | gaze point error (mm): 41.6 -> 38.7 | Improved |
| use refined depth for 3D eye position instead of raw depth (ablation) | Our method · ShanghaiTechGaze+ (RGBD) test | gaze point error (mm): 40.4 -> 38.7 | Improved |
| use accurate 3D eye positions instead of face grid (no-depth vs iTracker*) (ablation) | Our method · ShanghaiTechGaze+ (RGBD) test | gaze point error (mm): 47.5 -> 46.7 | Improved |
| add depth input to head pose branch (ablation) | Our method · EYEDIAP | gaze direction error (deg): 5.7 -> 4.8 | Improved |
| use ResNet-34 backbone in iTracker (iTracker*) | iTracker · ShanghaiTechGaze+ (RGBD) test | gaze point error (mm): 55.5 -> 47.5 | Improved |
| proposed RGBD multi-task method vs iTracker* | iTracker* · EYEDIAP | gaze direction error (deg): 5.7 -> 4.8 | Improved |
Closest papers by meaning
- RT-GENE: Real-Time Eye Gaze Estimation in Natural Environments (ECCV 2018 · similarity 0.88)
- Deep Pictorial Gaze Estimation (ECCV 2018 · similarity 0.88)
- Appearance-Based Gaze Estimation With Deep Learning: A Review and Benchmark (TPAMI 2024 · similarity 0.88)
- From Feature to Gaze: A Generalizable Replacement of Linear Layer for Gaze Estimation (CVPR 2024 · similarity 0.87)
- GazeGene: Large-scale Synthetic Gaze Dataset with 3D Eyeball Annotations (CVPR 2025 · similarity 0.87)
- GA3CE: Unconstrained 3D Gaze Estimation with Gaze-Aware 3D Context Encoding (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.
