Research map / Image generation / Editing and style transfer
Sparse, Smart Contours to Represent and Edit Images
Tali Dekel, Chuang Gan, Dilip Krishnan, Ce Liu, William T. Freeman
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 |
|---|---|---|---|
| replace a single GAN with the LFN-HFN cascade, using gradient features (ablation) | image reconstruction from sparse contours · VGG Faces | SSIM: 0.786 -> 0.812 | Improved |
| replace a single GAN with the LFN-HFN cascade, using gradient features (ablation) | image reconstruction from sparse contours · CUB Birds | SSIM: 0.697 -> 0.740 | Improved |
| replace a single GAN with the LFN-HFN cascade, using gradient features (ablation) | image reconstruction from sparse contours · Stanford Dogs | SSIM: 0.722 -> 0.749 | Improved |
| use learned features instead of gradient features (ablation) | LFN-HFN contour reconstruction · VGG Faces | SSIM: 0.812 -> 0.867 | Improved |
| use learned features instead of gradient features (ablation) | LFN-HFN contour reconstruction · CUB Birds | SSIM: 0.740 -> 0.783 | Improved |
| use learned features instead of gradient features (ablation) | LFN-HFN contour reconstruction · Stanford Dogs | SSIM: 0.749 -> 0.810 | Improved |
| use color features instead of gradient features (ablation) | LFN-HFN contour reconstruction · VGG Faces | SSIM: 0.812 -> 0.859 | Improved |
| use color features instead of gradient features (ablation) | LFN-HFN contour reconstruction · CUB Birds | SSIM: 0.740 -> 0.763 | Improved |
| use color features instead of gradient features (ablation) | LFN-HFN contour reconstruction · Stanford Dogs | SSIM: 0.749 -> 0.792 | Improved |
| use gradient features rather than learned features for contour edits (ablation) | LFN-HFN image editing · edited images | edit quality | Improved |
| use the dilated-patch discriminator instead of the patch discriminator (ablation) | LFN-HFN reconstruction · sparse contour reconstruction | reconstruction quality | Improved |
| add the HFN's adversarial training to the LFN reconstruction (ablation) | LFN-HFN reconstruction · VGG Faces | texture loss | Improved |
Closest papers by meaning
- Instance Contour Adjustment via Structure-Driven CNN (ECCV 2022 · similarity 0.87)
- The GAN That Warped: Semantic Attribute Editing With Unpaired Data (CVPR 2020 · similarity 0.86)
- Improving Shape Deformation in Unsupervised Image-to-Image Translation (ECCV 2018 · similarity 0.85)
- GANFIT: Generative Adversarial Network Fitting for High Fidelity 3D Face Reconstruction (CVPR 2019 · similarity 0.85)
- Deforming Autoencoders: Unsupervised Disentangling of Shape and Appearance (ECCV 2018 · similarity 0.85)
- In-N-Out: Faithful 3D GAN Inversion with Volumetric Decomposition for Face Editing (CVPR 2024 · 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.
