Research map / Image generation / Editing and style transfer
Quality Metric Guided Portrait Line Drawing Generation From Unpaired Training Data
Ran Yi, Yong‐Jin Liu, Yu-Kun Lai, Paul L. Rosin
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 relaxed cycle-consistency loss and local discriminators (ablation) | APDrawing generator · APDrawings, Style1 FID | FID: 102.9 -> 81.2 | Improved |
| add relaxed cycle-consistency loss (ablation) | APDrawing generator · APDrawings, Style1 FID | FID: 88.0 -> 81.2 | Improved |
| add local discriminators (ablation) | APDrawing generator · APDrawings, Style2 FID | FID: 142.3 -> 114.3 | Improved |
| add HED edge extraction in relaxed cycle loss (ablation) | APDrawing generator · APDrawings, Style3 FID | FID: 103.4 -> 89.7 | Improved |
| add quality loss from quality metric model (ablation) | APDrawing generator · APDrawings, Style3 FID | FID: 108.2 -> 89.7 | Improved |
| add style feature vector and style loss (ablation) | APDrawing generator · APDrawings, Style1 FID | FID: 114.3 -> 81.2 | Improved |
| add truncation loss (ablation) | APDrawing generator · APDrawings, Style1 FID | FID: 81.7 -> 81.2 | Improved |
| use local discriminators instead of single discriminator with face-region channel (ablation) | APDrawing generator · APDrawings, Style2 FID | FID: 152.8 -> 114.3 | Improved |
| use full method instead of conference version (Ours-pre, no relaxed/local/style improvements) | APDrawing generator · APDrawings, Style3 FID | FID: 108.2 -> 89.7 | Improved |
| use full method instead of DRIT | APDrawing generator · APDrawings, Style2 FID | FID: 135.0 -> 114.3 | Improved |
| use full method instead of ComboGAN in user study (ranked first) | APDrawing generator · User study | percentage ranked first: 23.7% -> 45.1% | Improved |
| use full method instead of Ours-pre on quality metric score | APDrawing generator · APDrawings test set | Quality score: 0.45 -> 0.51 | 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.
- DRIT — DRIT++: Diverse Image-to-Image Translation via Disentangled Representations (IJCV 2020) · better on 4 of 4 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
- Unpaired Portrait Drawing Generation via Asymmetric Cycle Mapping (CVPR 2020 · similarity 0.94)
- Learning Portrait Drawing with Unsupervised Parts (IJCV 2023 · similarity 0.91)
- Learning To Generate Line Drawings That Convey Geometry and Semantics (CVPR 2022 · similarity 0.89)
- Line Drawings for Face Portraits From Photos Using Global and Local Structure Based GANs (TPAMI 2020 · similarity 0.88)
- LineArt: A Knowledge-guided Training-free High-quality Appearance Transfer for Design Drawing with Diffusion Model (CVPR 2025 · similarity 0.84)
- ID-Sculpt: ID-aware 3D Head Generation from Single In-the-wild Portrait Image (AAAI 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.
