Research map / Label-efficient and robust learning / Recognition and long-tail
A Comprehensive Performance Evaluation of Deformable Face Tracking “In-the-Wild”
Grigorios G. Chrysos, Epameinondas Antonakos, Patrick Snape, Akshay Asthana, Stefanos P. Zafeiriou
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 reinitialisation from previous frame on missed detections (ablation) | DPM + AAM · 300VW Category 1 | AUC: 0.447 -> 0.572 | Improved |
| add reinitialisation from previous frame on missed detections (ablation) | SS-DPM + AAM · 300VW Category 3 | AUC: 0.161 -> 0.175 | Improved |
| add reinitialisation from previous frame on missed detections (ablation) | SVM-HOG + AAM · 300VW Category 1 | AUC: 0.493 -> 0.627 | Improved |
| add reinitialisation from previous frame on missed detections (ablation) | DPM + CFSS · 300VW Category 1 | AUC: 0.764 -> 0.765 | No clear effect |
| add reinitialisation from previous frame on missed detections (ablation) | DPM + ERT · 300VW Category 1 | AUC: 0.772 -> 0.773 | No clear effect |
| use CFSS instead of SDM landmark localisation | DPM detector · 300VW Category 1 | AUC: 0.673 -> 0.764 | Improved |
| use ERT instead of SDM landmark localisation | DPM detector · 300VW Category 1 | AUC: 0.673 -> 0.772 | Improved |
| use AAM instead of CFSS landmark localisation | DPM detector · 300VW Category 3 | AUC: 0.717 -> 0.376 | Hurt |
| use weakly supervised DPM instead of strongly supervised SS-DPM detector | ERT landmark · 300VW Category 3 | AUC: 0.243 -> 0.714 | Improved |
| use DPM instead of VJ detector | CFSS landmark · 300VW Category 1 | AUC: 0.660 -> 0.764 | Improved |
| use SPOT tracker instead of previous-frame initialisation (PREV) | CFSS landmark · 300VW Category 1 | AUC: 0.545 -> 0.769 | Improved |
| use SRDCF tracker instead of PREV | CFSS landmark · 300VW Category 3 | AUC: 0.199 -> 0.687 | Improved |
Closest papers by meaning
- PD 2 T: Person-Specific Detection, Deformable Tracking (TPAMI 2017 · similarity 0.87)
- Face Landmark Fitting via Optimized Part Mixtures and Cascaded Deformable Model (TPAMI 2016 · similarity 0.86)
- A Functional Regression Approach to Facial Landmark Tracking (TPAMI 2017 · similarity 0.86)
- Synergy Between Face Alignment and Tracking via Discriminative Global Consensus Optimization (ICCV 2017 · similarity 0.85)
- 3D Face Tracking from 2D Video through Iterative Dense UV to Image Flow (CVPR 2024 · similarity 0.84)
- Is This Tracker On? A Benchmark Protocol for Dynamic Tracking (NeurIPS 2025 · similarity 0.82)
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.
