Research map / 3D vision / Human pose estimation
Multi-Task Head Pose Estimation in-the-Wild
Roberto Valle, José M. Buenaposada, Luis Baumela
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 |
|---|---|---|---|
| attach pose losses at the encoder bottleneck instead of the decoder end (ablation) | MNN · Overall validation loss | loss: 18.6 -> 7.9 | Improved |
| add landmark pre-training before multi-task training (ablation) | MNN · Overall validation loss | loss: 8.7 -> 7.9 | Improved |
| use symmetric multi-task learning instead of single-task pose (ablation) | MNN · Head pose avg (5 sets) | mean MAE (deg): 2.54 -> 2.35 | Improved |
| add landmark pre-training to multi-task learning (ablation) | MNN · Head pose avg (5 sets) | mean MAE (deg): 2.35 -> 2.24 | Improved |
| fine-tune only on pose instead of all three tasks after pre-training (ablation) | MNN · Head pose avg (5 sets) | mean MAE (deg): 2.24 -> 2.24 | No clear effect |
| add multi-task learning to visibility estimation (ablation) | MNN · COFW visibility | recall at 80% precision (%): 21.75 -> 72.12 | Improved |
| add occlusion-aware regressor (OR) (ablation) | MNN · COFW | NME: 5.65 -> 5.04 | Improved |
| add occlusion-aware regressor (OR) (ablation) | MNN · AFLW | NME: 2.03 -> 1.97 | Improved |
| add occlusion-aware regressor (OR) (ablation) | MNN · AFLW2000-3D | NME: 2.77 -> 2.58 | Improved |
| use corrected pose annotations (AFLW2000-3D-POSIT) for evaluation | MNN · AFLW2000-3D | mean MAE (deg): 3.83 -> 1.71 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- HopeNet — HOPE-Net: A Graph-Based Model for Hand-Object Pose Estimation (CVPR 2020) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Pose-Invariant Face Alignment With a Single CNN (ICCV 2017 · similarity 0.88)
- HyperFace: A Deep Multi-Task Learning Framework for Face Detection, Landmark Localization, Pose Estimation, and Gender Recognition (TPAMI 2017 · similarity 0.88)
- Binarized Convolutional Landmark Localizers for Human Pose Estimation and Face Alignment With Limited Resources (ICCV 2017 · similarity 0.88)
- Learning Pose-Aware Models for Pose-Invariant Face Recognition in the Wild (TPAMI 2018 · similarity 0.87)
- TokenHPE: Learning Orientation Tokens for Efficient Head Pose Estimation via Transformers (CVPR 2023 · similarity 0.87)
- Pose-Dependent Low-Rank Embedding for Head Pose Estimation (AAAI 2016 · 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.
