Research map / 3D vision / Depth and stereo
Colorization of Depth Map via Disentanglement
Chung-Sheng Lai, Zunzhi You, Ching-Chun Huang, Yi-Hsuan Tsai, Wei-Chen Chiu
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 random flipping of the reference pair (ablation) | Our model · NYU Depth v2 | FID: 58.9894 -> 45.1402 | Improved |
| add random flipping of the reference pair (ablation) | Our model · NYU Depth v2 | PSNR: 13.8649 -> 12.1115 | Hurt |
| add time invariant property (TIP) supervision (ablation) | Our model · NYU Depth v2 | FID: 138.9675 -> 45.1402 | Improved |
| add time invariant property (TIP) supervision (ablation) | Our model · NYU Depth v2 | PSNR: 16.1137 -> 12.1115 | Hurt |
| add DRIT-arch cycle consistency and discriminator (ablation) | Our model · NYU Depth v2 | FID: 55.9795 -> 45.1402 | Improved |
| add DRIT-arch cycle consistency and discriminator (ablation) | Our model · NYU Depth v2 | PSNR: 12.7394 -> 12.1115 | Hurt |
| use disentangled appearance/structure model instead of BicycleGAN | BicycleGAN · NYU Depth v2 | FID: 145.3382 -> 45.1402 | Improved |
| use disentangled model instead of Pix2Pix | Pix2Pix · SceneNet RGB-D | FID: 152.3537 -> 92.7267 | Improved |
| use disentangled model instead of BicycleGAN | BicycleGAN · SceneNet RGB-D | PSNR: 15.5830 -> 22.3333 | Improved |
| use colorized depth from our model instead of Pix2Pix for YOLOv3 detection | Pix2Pix · NYU Depth v2 chair detection consistency | AP: 38.0 -> 69.9 | Improved |
| use a different appearance reference instead of the original image's own pair | Our Full Model · NYU Depth v2 chair detection consistency | AP: 71.0 -> 69.9 | No clear effect |
| colorize with our model instead of leaving the image ill-lighted | Ill-Lighted · NYU Depth v2 chair detection consistency | AP: 3.5 -> 69.9 | Improved |
Closest papers by meaning
- Perceptual Deep Depth Super-Resolution (ICCV 2019 · similarity 0.85)
- RGB-D Local Implicit Function for Depth Completion of Transparent Objects (CVPR 2021 · similarity 0.85)
- Neural RGB(r)D Sensing: Depth and Uncertainty From a Video Camera (CVPR 2019 · similarity 0.85)
- Guest Editors’ Introduction to the Special Issue on RGB-D Vision: Methods and Applications (TPAMI 2020 · similarity 0.85)
- Masked Depth Modeling for Spatial Perception (ECCV 2026 · similarity 0.84)
- Depth Prompting for Sensor-Agnostic Depth Estimation (CVPR 2024 · similarity 0.83)
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.
