Research map / 3D vision / Human pose estimation
Binarized Convolutional Landmark Localizers for Human Pose Estimation and Face Alignment With Limited Resources
Adrian Bulat, Georgios Tzimiropoulos
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 |
|---|---|---|---|
| replace bottleneck with proposed hierarchical parallel multi-scale block at same parameters | binary Hourglass · MPII val (binary) | PCKh: 67.2% -> 72.7% | Improved |
| replace bottleneck with proposed block (6.2M vs wider 7.0M) in real-valued network | real Hourglass · MPII val (real) | PCKh: 83.1% -> 85.5% | Improved |
| widen thin 3x3 layer from 128 to 256 channels (ablation) | binary bottleneck · MPII val (binary) | PCKh: 67.2% -> 70.7% | Improved |
| use multi-scale filters (ablation) | binary bottleneck · MPII val (binary) | PCKh: 67.2% -> 69.3% | Improved |
| remove 1x1 convolutions from multi-scale block (ablation) | binary MS block · MPII val (binary) | PCKh: 69.3% -> 75.5% | Improved |
| add hierarchical parallel structure for gradient flow (ablation) | binary MS block without 1x1 · MPII val (binary) | PCKh: 75.5% -> 76% | Improved |
| add data augmentation (ablation) | binary final model · MPII val (binary) | PCKh: 72.1% -> 76% | Improved |
| use sigmoid cross-entropy loss instead of L2 (ablation) | binary final model · MPII val (binary) | PCKh: 73.8% -> 76% | Improved |
| use max pooling instead of average pooling (ablation) | binary final model · MPII val (binary) | PCKh: 71.9% -> 76% | Improved |
| add ReLU after each convolutional layer (ablation) | binary final model · MPII val (binary) | PCKh: 76% -> 77.8% | Improved |
| binarize the bottleneck Hourglass | Hourglass · MPII val | PCKh: 76.5 -> 67.2 | Hurt |
| use proposed binary block in segmentation instead of binary bottleneck | binary network · 300W part segmentation | mean IU: 62.49% -> 68.05% | 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.
- HyperFace — HyperFace: A Deep Multi-Task Learning Framework for Face Detection, Landmark Localization, Pose Estimation, and Gender Recognition (TPAMI 2017) · better on 1 of 1 shared result
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
- Hierarchical Binary CNNs for Landmark Localization with Limited Resources (TPAMI 2018 · similarity 0.98)
- Adversarial Learning of Structure-Aware Fully Convolutional Networks for Landmark Localization (TPAMI 2019 · similarity 0.89)
- Facial Landmark Detection with Tweaked Convolutional Neural Networks (TPAMI 2017 · similarity 0.88)
- Disentangling 3D Pose in a Dendritic CNN for Unconstrained 2D Face Alignment (CVPR 2018 · similarity 0.88)
- Multi-Task Head Pose Estimation in-the-Wild (TPAMI 2020 · similarity 0.88)
- Neural Localizer Fields for Continuous 3D Human Pose and Shape Estimation (NeurIPS 2024 · 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.
