Research map / Detection and segmentation / Weakly supervised detection and segmentation
Teacher Supervises Students How to Learn From Partially Labeled Images for Facial Landmark Detection
Xuanyi Dong, Yi Yang
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 the HG student and student interaction to a single CPM student (10% labels) (ablation) | TS3 · 300-W Full test | NME: 8.28 -> 5.64 | Improved |
| add the CPM student and student interaction to a single HG student (10% labels) (ablation) | TS3 · 300-W Full test | NME: 6.25 -> 5.64 | Improved |
| add the HG student to a single CPM student (20% labels) (ablation) | TS3 · 300-W Full test | NME: 6.68 -> 5.03 | Improved |
| add the CPM student to a single HG student (20% labels) (ablation) | TS3 · 300-W Full test | NME: 6.68 -> 5.03 | Improved |
| add the HG student to CPM on challenging subset (10% labels) (ablation) | TS3 · 300-W Challenging | NME: 14.69 -> 9.26 | Improved |
| add the learned teacher to select pseudo labels, versus SPL and SPaCo selection (ablation) | TS3 · 300-W Full test (10% labels) | NME | Improved |
| raise label ratio from 10% to 20% | TS3 · 300-W Full test | NME: 5.64 -> 5.03 | Improved |
| use TS3 instead of RCN+ with 10% labels | TS3 · 300-W Full test | NME: 6.32 -> 5.64 | Improved |
| use TS3 instead of RCN+ with 20% labels | TS3 · 300-W Full test | NME: 5.88 -> 5.03 | Improved |
| add unlabeled AFLW training data to 100% labeled 300-W | TS3 · 300-W Full test | NME: 3.78 -> 3.49 | Improved |
| increase label ratio from 5% to 20% | TS3 · AFLW-Full | NME: 2.19 -> 1.99 | Improved |
| add unlabeled 300-VW videos with only 300-W labels, versus fully labeled baselines | TS3 · 300-VW category C | AUC@0.08: 59.39 -> 59.65 | Improved |
Closest papers by meaning
- Improving Landmark Localization With Semi-Supervised Learning (CVPR 2018 · similarity 0.88)
- Brute-Force Facial Landmark Analysis With a 140,000-Way Classifier (AAAI 2018 · similarity 0.87)
- Humble Teachers Teach Better Students for Semi-Supervised Object Detection (CVPR 2021 · similarity 0.87)
- Unbiased Teacher for Semi-Supervised Object Detection (ICLR 2021 · similarity 0.87)
- Aggregation via Separation: Boosting Facial Landmark Detector With Semi-Supervised Style Translation (ICCV 2019 · similarity 0.87)
- LAFS: Landmark-based Facial Self-supervised Learning for Face Recognition (CVPR 2024 · similarity 0.85)
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.
