Research map / Image generation / Editing and style transfer
Semantic-Sparse Colorization Network for Deep Exemplar-Based Colorization
Yunpeng Bai, Chao Dong, Zenghao Chai, Andong Wang, Zhengzhuo Xu, Chun Yuan
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 the LDT fine colorization stage after coarse GCT stage (ablation) | SSCN · Self-augmentation PSNR/SSIM | PSNR/SSIM: 30.02 -> 34.73 PSNR; 0.937 -> 0.964 SSIM | Improved |
| add GCT coarse colorization to build correspondence instead of gray-scale features (ablation) | SSCN · Self-augmentation PSNR/SSIM | PSNR/SSIM: PSNR and SSIM decrease by 3.30 and 0.014 when removed | Improved |
| add classification loss (ablation) | SSCN dense version · Self-augmentation PSNR/SSIM | PSNR/SSIM: 33.73 -> 35.25 PSNR; 0.952 -> 0.961 SSIM | Improved |
| add color histogram loss (ablation) | SSCN (k=256, r=256) · Self-augmentation PSNR/SSIM | PSNR/SSIM: PSNR and SSIM decrease by 1.28 and 0.009 when removed | Improved |
| add random background selection S_r (r>0) (ablation) | SSCN · Self-augmentation (TPS reference) | PSNR/SSIM | Improved |
| increase r further beyond the best value (ablation) | SSCN · Self-augmentation (TPS reference) | PSNR/SSIM | Hurt |
| use larger or smaller top-k than 256 (ablation) | SSCN · Self-augmentation (TPS reference) | PSNR/SSIM | Hurt |
| replace dense attention with semantic-sparse selection (k=256, r=256) | SSCN · Self-augmentation (TPS reference) | PSNR: 35.25 -> 36.32 PSNR; 0.961 -> 0.969 SSIM | Improved |
Closest papers by meaning
- Stylization-Based Architecture for Fast Deep Exemplar Colorization (CVPR 2020 · similarity 0.90)
- BiSTNet: Semantic Image Prior Guided Bidirectional Temporal Feature Fusion for Deep Exemplar-Based Video Colorization (TPAMI 2024 · similarity 0.89)
- Deep Exemplar-Based Video Colorization (CVPR 2019 · similarity 0.89)
- LGA-Net: Learning Local and Global Affinities for Sparse Scribble based Image Colorization (ICCV 2025 · similarity 0.89)
- Pixelated Semantic Colorization (IJCV 2019 · similarity 0.89)
- ColorFormer: Image Colorization via Color Memory Assisted Hybrid-Attention Transformer (ECCV 2022 · similarity 0.88)
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.
