Research map / 3D vision / Depth and stereo
Rendering Portraitures from Monocular Camera and Beyond
Xiangyu Xu, Deqing Sun, Sifei Liu, Wenqi Ren, Yu-Jin Zhang, Ming-Hsuan Yang, Jian Sun
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 SPN refinement to the coarse portrait segmentation | portrait segmentation · Baidu human segmentation (whole-body) | IoU: 91.43 -> 93.22 | Improved |
| add SPN refinement to selfie segmentation | portrait segmentation · Selfie segmentation dataset | mean IoU (%): 95.52 -> 96.40 | Improved |
| refine depth with SPN trained on sparse ranking labels | depth estimation · Depth in the Wild (DIW) test | WHDR: 14.39 -> 14.35 | No clear effect |
| add the body depth-consistency loss with portrait segmentation data (ablation) | SPN depth estimation · portrait depth qualitative | depth consistency on human body | Improved |
| use spatially-variant RNN with depth and segmentation guidance instead of CNN (ablation) | DoF approximation network · Baidu human test set (DoF approximation) | PSNR (dB): 37.35 -> 40.74 | Improved |
| use spatially-variant RNN with depth and segmentation guidance instead of CNN (ablation) | DoF approximation network · Baidu human test set (DoF approximation) | SSIM: 0.9723 -> 0.9868 | Improved |
| add depth and segmentation guidance to RNN filtering compared with Liu et al. | RNN filter · Baidu human test set (DoF approximation) | PSNR (dB): 33.55 -> 40.74 | Improved |
| add depth layers to background blur (ablation) | CRF-based DoF rendering · User study (22 subjects, 30 whole-body images) | percent judged real SLR: 13.2% -> 79.1% | Improved |
| add depth information to the blur (ablation) | CRF-based DoF rendering · qualitative ablation | visual realism | Improved |
| add SPN refinement of segmentation (ablation) | CRF-based DoF rendering · qualitative ablation | visual realism | Improved |
| use CRF depth layering instead of naive thresholding (ablation) | CRF-based DoF rendering · qualitative ablation | visual realism | Improved |
| add segmentation-guided blur kernel (ablation) | CRF-based DoF rendering · qualitative ablation | boundary artifacts | Improved |
Closest papers by meaning
- Learning Depth from Focus in the Wild (ECCV 2022 · similarity 0.88)
- Learning Depth from Single Monocular Images Using Deep Convolutional Neural Fields (TPAMI 2016 · similarity 0.88)
- User-Instructed Disparity-aware Defocus Control (NeurIPS 2025 · similarity 0.87)
- Learning Single Camera Depth Estimation Using Dual-Pixels (ICCV 2019 · similarity 0.87)
- Lens Parameter Estimation for Realistic Depth of Field Modeling (ICCV 2023 · similarity 0.86)
- DepthFM: Fast Generative Monocular Depth Estimation with Flow Matching (AAAI 2025 · similarity 0.86)
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.
