Research map / Image generation / Editing and style transfer
DanbooRegion: An Illustration Region Dataset
Lvmin Zhang, Yi JI, Chunping Liu
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 FAR assignment recommendation to the annotation workflow (ablation) | annotation workflow, artist #1 · Annotation workflow | wasted time share: 53m (70.13%) -> 10m (13.15%) | Improved |
| add FAR assignment recommendation to the annotation workflow (ablation) | annotation workflow, artist #2 · Annotation workflow | wasted time share: 46m (58.97%) -> 7m (11.47%) | Improved |
| add FAR assignment recommendation to the annotation workflow (ablation) | annotation workflow, artist #3 · Annotation workflow | wasted time share: 37m (51.39%) -> 5m (6.67%) | Improved |
| use deeper encoder VGG19 instead of VGG16 | U-net region-from-normal · DanbooRegion test set (1000) | ODS: 0.731 -> 0.647 | Hurt |
| use deeper encoder Resnet110 instead of Resnet56 | U-net region-from-normal · DanbooRegion test set (1000) | ODS: 0.732 -> 0.692 | Hurt |
| use deeper encoder Resnet110 instead of Resnet56 | U-net region-from-normal · DanbooRegion test set (1000) | AP: 0.722 -> 0.641 | Hurt |
| use learning-based Pix2PixHD instead of best traditional Quad-Tree/Felz-Hutt segmentation | region extraction · DanbooRegion test set (1000) | ODS: 0.466 -> 0.57 | Improved |
| use learning-based VGG16 U-net instead of traditional gPb-owt-ucm | region extraction · DanbooRegion test set (1000) | ODS: 0.640 -> 0.731 | Mixed |
| use instance segmentation TensorMask instead of VGG16 U-net region-from-normal | region extraction · DanbooRegion test set (1000) | ODS: 0.731 -> 0.495 | Hurt |
| use region-from-normal VGG16 U-net instead of image-to-image Pix2Pix | region extraction · DanbooRegion test set (1000) | AP: 0.562 -> 0.710 | Improved |
| use best model versus human re-annotation | Resnet56 U-net · DanbooRegion test set (1000) | ODS: 0.732 -> 0.782 | Hurt |
Closest papers by meaning
- Learning Inclusion Matching for Animation Paint Bucket Colorization (CVPR 2024 · similarity 0.84)
- Advancing Manga Analysis: Comprehensive Segmentation Annotations for the Manga109 Dataset (CVPR 2025 · similarity 0.83)
- Prune and Repaint: Content-Aware Image Retargeting for any Ratio (NeurIPS 2024 · similarity 0.83)
- Image Referenced Sketch Colorization Based on Animation Creation Workflow (CVPR 2025 · similarity 0.83)
- AniDoc: Animation Creation Made Easier (CVPR 2025 · similarity 0.83)
- APISR: Anime Production Inspired Real-World Anime Super-Resolution (CVPR 2024 · similarity 0.83)
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.
