Research map / 3D vision / Face and shape
Learning Perspective Undistortion of Portraits
Yajie Zhao, Zeng Huang, Tianye Li, Weikai Chen, Chloe LeGendre, Xinglei Ren, Ari Shapiro, Hao Li
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 Fried et al. with learned cascade undistortion | portrait undistortion · BU-4DFE synthetic pairs | average intensity error: 1.28 -> 0.39 | Improved |
| add FlowNet before CompletionNet (cascade) vs single image-to-image network (ablation) | portrait undistortion · qualitative ablation | image sharpness | Improved |
| add distance label channel as input to FlowNet (ablation) | FlowNet · qualitative ablation | distortion of nose, mouth, face rim | Improved |
| undistort near-range portrait before face verification | OpenFace · BU-4DFE face verification | accuracy: 0.9137 -> 0.9473 | Improved |
| undistort portrait before landmark detection | OpenPose · BU-4DFE renderings | detection failure rate: 9.0% -> 0% | Improved |
| undistort portrait before landmark detection | OpenPose · BU-4DFE renderings | NME: 4.9% -> 4.4% | Improved |
| undistort portrait before 3D face reconstruction | FaceWarehouse fitting · near-range portraits | reconstruction error | Improved |
Closest papers by meaning
- Practical Wide-Angle Portraits Correction With Deep Structured Models (CVPR 2021 · similarity 0.92)
- DisCO: Portrait Distortion Correction with Perspective-Aware 3D GANs (IJCV 2024 · similarity 0.91)
- Combining Generative and Geometry Priors for Wide-Angle Portrait Correction (ECCV 2024 · similarity 0.87)
- An End-to-End Depth-Based Pipeline for Selfie Image Rectification (TPAMI 2025 · similarity 0.87)
- DiffPortrait3D: Controllable Diffusion for Zero-Shot Portrait View Synthesis (CVPR 2024 · similarity 0.84)
- MaDCoW: Marginal Distortion Correction for Wide-Angle Photography with Arbitrary Objects (CVPR 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.
