Research map / Image generation / Editing and style transfer
Tag2Pix: Line Art Colorization Using Text Tag With SECat and Changing Loss
Hyunsu Kim, Ho Young Jhoo, Eunhyeok Park, Sungjoo Yoo
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 |
|---|---|---|---|
| use SECat block instead of plain ResNeXt block (ablation) | Tag2Pix generator · CVT embedding comparison | FID: 52.49 -> 39.21 | Improved |
| use SECat instead of SE-ResNeXt block (ablation) | Tag2Pix generator · CVT embedding comparison | FID: 45.48 -> 39.21 | Improved |
| use SECat instead of concatenating CVT before the first conv (ablation) | Tag2Pix generator · CVT embedding comparison | FID: 39.69 -> 39.21 | Improved |
| use SECat instead of concatenating CVT before every conv (ablation) | Tag2Pix generator · CVT embedding comparison | FID: 47.08 -> 39.21 | Improved |
| use SECat instead of AdaIN (ablation) | Tag2Pix generator · CVT embedding comparison | FID: 66.39 -> 39.21 | Improved |
| use SECat instead of concatenating CVT front (ablation) | Tag2Pix generator · CVT embedding comparison | parameters: 16.03M -> 15.81M | Improved |
| use SECat instead of SE-ResNeXt (ablation) | Tag2Pix generator · User study of CVT embedding | Quality: 2.78 -> 3.54 | Improved |
| use SECat instead of concat front (ablation) | Tag2Pix generator · User study of CVT embedding | Segmentation: 3.29 -> 3.51 | Improved |
| use Tag2Pix instead of Style2Paints | Tag2Pix · Sketch-based user study | Quality: 3.47 -> 3.86 | Improved |
| use Tag2Pix instead of PaintsChainer | Tag2Pix · Sketch-based user study | Segmentation: 2.51 -> 3.94 | Improved |
| use Tag2Pix instead of Manjunatha et al. | Tag2Pix · Text-based user study | Segmentation: 3.16 -> 4.13 | Improved |
| use two-step curriculum training (segmentation then colorization) instead of single-step (ablation) | Tag2Pix · line art colorization | color mixing and bleeding | Improved |
Closest papers by meaning
- Style-Structure Disentangled Features and Normalizing Flows for Diverse Icon Colorization (CVPR 2022 · similarity 0.87)
- Eliminating Gradient Conflict in Reference-Based Line-Art Colorization (ECCV 2022 · similarity 0.86)
- MangaNinja: Line Art Colorization with Precise Reference Following (CVPR 2025 · similarity 0.85)
- Automatic Controllable Colorization via Imagination (CVPR 2024 · similarity 0.83)
- Free-Lunch Color-Texture Disentanglement for Stylized Image Generation (NeurIPS 2025 · similarity 0.83)
- Learning Inclusion Matching for Animation Paint Bucket Colorization (CVPR 2024 · 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.
