Research map / Label-efficient and robust learning / Recognition and long-tail
Face Parsing With RoI Tanh-Warping
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Train on relabeled HELEN* rather than HELEN | Face Parsing with RoI Tanh-Warping · HELEN test | overall F-measure: 92.4 -> 93.1 | Helped |
| Train on relabeled HELEN* rather than HELEN | Face Parsing with RoI Tanh-Warping · HELEN test | hair F-measure: 83.5 -> 88.7 | Helped |
| Use RoI Tanh-warping instead of face-rectangle cropping (ablation) | Hybrid face parsing network · HELEN* | hair F-measure: 84.5 -> 88.7 | Helped |
| Use RoI Tanh-warping instead of face-rectangle cropping (ablation) | Hybrid face parsing network · HELEN* | overall F-measure: 92.9 -> 93.1 | Helped |
| Replace the hybrid inner/outer branches with FCN (ablation) | Tanh-warped face parsing network · HELEN* | overall F-measure: 89.3 -> 93.1 | Hurt |
| Add the hybrid component branches rather than using FCN alone (ablation) | Tanh-warped face parsing network · HELEN* | hair F-measure: 88.8 vs. 88.7 | No clear effect |
| Add padding to regressed component boxes (ablation) | Hybrid face parsing network · HELEN* | overall F-measure: 91.9 -> 93.1 | Helped |
| Keep separate weights for inner-component segmentation modules instead of sharing them (ablation) | Hybrid face parsing network · HELEN* | overall F-measure: 92.7 -> 93.1 | Helped |
| Share weights between left/right eye and brow branches with horizontal flips (ablation) | Hybrid face parsing network · HELEN* | overall F-measure: 93.0 vs. 93.1 | No clear effect |
| Double the face crop area (ablation) | Hybrid face parsing network · HELEN* | overall F-measure: 92.9 -> 90.9 | Hurt |
| Double the face crop area (ablation) | Hybrid face parsing network · HELEN* | hair F-measure: 84.5 -> 83.8 | Hurt |
| Use face-focused crop or Tanh-warping instead of rescaling the full image (ablation) | FCN face parsing network · HELEN* | overall F-measure: 73.0 -> 89.3 | Helped |
Closest papers by meaning
- SegFace: Face Segmentation of Long-Tail Classes (AAAI 2025 · similarity 0.88)
- A New Dataset and Boundary-Attention Semantic Segmentation for Face Parsing (AAAI 2020 · similarity 0.88)
- GroundingFace: Fine-grained Face Understanding via Pixel Grounding Multimodal Large Language Model (CVPR 2025 · similarity 0.85)
- Neural Implicit Morphing of Face Images (CVPR 2024 · similarity 0.85)
- Self-Supervised Facial Representation Learning with Facial Region Awareness (CVPR 2024 · similarity 0.85)
- FaceBench: A Multi-View Multi-Level Facial Attribute VQA Dataset for Benchmarking Face Perception MLLMs (CVPR 2025 · similarity 0.84)
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.