Research map / 3D vision / Depth and stereo
Du²Net: Learning Depth Estimation from Dual-Cameras and Dual-Pixels
Yinda Zhang, Neal Wadhwa, Sergio Orts-Escolano, Christian Häne, Sean Fanello, Rahul Garg
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 dual-pixel data to the dual-camera confidence volume (ablation) | Du2Net unrefined disparity · Volume fusion ablation (all pixels) | MAE: 1.023 -> 0.889 | Improved |
| add dual-pixel data to the dual-camera confidence volume (ablation) | Du2Net unrefined disparity · Volume fusion ablation (all pixels) | bad delta>2: 10.65 -> 8.78 | Improved |
| fuse confidence volumes instead of 2D disparity maps (ablation) | Du2Net unrefined disparity · Volume fusion ablation (all pixels) | MAE: 0.969 -> 0.889 | Improved |
| fuse confidence volumes instead of cost volumes (ablation) | Du2Net unrefined disparity · Volume fusion ablation (all pixels) | MAE: 0.964 -> 0.889 | Improved |
| fuse DP with DC volume (ablation) | Du2Net unrefined disparity · Volume fusion ablation (occluded pixels) | MAE: 3.956 -> 3.526 | Improved |
| add DP to RGB as refinement guide (ablation) | Du2Net refinement · Refinement ablation (all pixels) | MAE: 0.838 -> 0.802 | Improved |
| add RGB to DP-only refinement guide (ablation) | Du2Net refinement · Refinement ablation (all pixels) | MAE: 0.829 -> 0.802 | Improved |
| use unwarped DP features for refinement instead of warped DP images (ablation) | Du2Net refinement · Refinement ablation (all pixels) | MAE: 0.835 -> 0.802 | Improved |
| add DP to RGB-guided refinement (ablation) | Du2Net refinement · Refinement ablation (all pixels) | bad delta>2: 7.74 -> 7.17 | Improved |
| use Du2Net (DC+DP) instead of PSM-Net (DC) | Du2Net · Dual-camera + dual-pixel test, all pixels | MAE: 0.815 -> 0.802 | Improved |
| use Du2Net (DC+DP) instead of PSM-Net (DC) | Du2Net · Dual-camera + dual-pixel test, occluded pixels | MAE: 2.799 -> 2.396 | Improved |
| use Du2Net (DC+DP) instead of DPNet (DP) | Du2Net* · Dual-camera + dual-pixel test, all pixels | MAE: 1.090 -> 0.746 | Improved |
Closest papers by meaning
- Learning Single Camera Depth Estimation Using Dual-Pixels (ICCV 2019 · similarity 0.93)
- On the Synergies between Machine Learning and Binocular Stereo for Depth Estimation from Images: a Survey (TPAMI 2021 · similarity 0.91)
- A Survey on Deep Learning Techniques for Stereo-Based Depth Estimation (TPAMI 2020 · similarity 0.91)
- Weakly-Supervised Depth Estimation and Image Deblurring via Dual-Pixel Sensors (TPAMI 2024 · similarity 0.90)
- Two-in-One Depth: Bridging the Gap Between Monocular and Binocular Self-Supervised Depth Estimation (ICCV 2023 · similarity 0.90)
- StereoNet: Guided Hierarchical Refinement for Real-Time Edge-Aware Depth Prediction (ECCV 2018 · similarity 0.90)
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.
