Research map / Image generation / Editing and style transfer
Instance Contour Adjustment via Structure-Driven CNN
Shuchen Weng, Yi Wei, Ming-Ching Chang, Boxin Shi
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 appearance affinity to SD-Conv ranking (ablation) | structure-driven CNN · Dilation | SSIM: 95.37 -> 95.48 | Improved |
| add geometry affinity to SD-Conv ranking (ablation) | structure-driven CNN · Dilation | SSIM: 95.39 -> 95.48 | Improved |
| add appearance affinity to SD-Conv ranking (ablation) | structure-driven CNN · Erosion | SSIM: 95.46 -> 95.53 | Improved |
| add geometry affinity to SD-Conv ranking (ablation) | structure-driven CNN · Erosion | SSIM: 95.42 -> 95.53 | Improved |
| use SD-Conv instead of Gated-Conv (ablation) | structure-driven CNN · Dilation | FID: 12.77 -> 12.43 | Improved |
| use SD-Conv instead of Deform-Conv (ablation) | structure-driven CNN · Dilation | FID: 12.99 -> 12.43 | Improved |
| use SD-Conv instead of Gated-Conv (ablation) | structure-driven CNN · Erosion | PSNR: 29.35 -> 29.38 | Mixed |
| add geometry affinity to SD-Attn (ablation) | structure-driven CNN · Dilation | SSIM: 95.40 -> 95.48 | Improved |
| add geometry affinity to SD-Attn (ablation) | structure-driven CNN · Dilation | PSNR: 28.29 -> 28.32 | No clear effect |
| use SD-Attn instead of contextual attention (ablation) | structure-driven CNN · Dilation | FID: 13.99 -> 12.43 | Improved |
| use SD-Attn instead of contextual attention (ablation) | structure-driven CNN · Dilation | PSNR: 27.52 -> 28.32 | Improved |
| use SD-Attn instead of contextual attention (ablation) | structure-driven CNN · Erosion | FID: 13.30 -> 11.64 | Improved |
Closest papers by meaning
- Sparse, Smart Contours to Represent and Edit Images (CVPR 2018 · similarity 0.87)
- DeFLOCNet: Deep Image Editing via Flexible Low-Level Controls (CVPR 2021 · similarity 0.86)
- Improving Editability in Image Generation with Layer-wise Memory (CVPR 2025 · similarity 0.86)
- ImgEdit: A Unified Image Editing Dataset and Benchmark (NeurIPS 2025 · similarity 0.84)
- PAIR Diffusion: A Comprehensive Multimodal Object-Level Image Editor (CVPR 2024 · similarity 0.84)
- ImageNet-trained CNNs are not biased towards texture: Revisiting feature reliance through controlled suppression (NeurIPS 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.
