Research map / Detection and segmentation / 3D detection and driving
Deep Optics for Monocular Depth Estimation and 3D Object Detection
Julie Chang, Gordon Wetzstein
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 defocus-coded optics instead of all-in-focus images | U-Net depth estimator · NYU Depth v2 | RMSE_lin: 0.9556 -> 0.4814 | Improved |
| add astigmatism to the defocus lens (ablation) | U-Net depth estimator · NYU Depth v2 | RMSE_lin: 0.4814 -> 0.4561 | Improved |
| add chromatic aberration of a singlet lens (ablation) | U-Net depth estimator · NYU Depth v2 | RMSE_lin: 0.4561 -> 0.4496 | Improved |
| end-to-end optimize a freeform lens instead of fixed chromatic aberration (ablation) | U-Net depth estimator · KITTI subset | RMSE_lin: 2.2566 -> 1.9288 | Improved |
| end-to-end optimize a freeform lens instead of fixed chromatic aberration (ablation) | U-Net depth estimator · Rectangles | RMSE_lin: 0.0984 -> 0.0902 | Improved |
| optimize annular lens instead of fixed chromatic aberration (ablation) | U-Net depth estimator · NYU Depth v2 | RMSE_lin: 0.4496 -> 0.4817 | Hurt |
| optimize annular lens instead of all-in-focus images | U-Net depth estimator · KITTI subset | RMSE_lin: 2.9100 -> 2.7998 | Improved |
| use freeform optimized lens on KITTI object detection | Frustum PointNet · KITTI object detection (val split) | 2D mAP: 78.01 -> 78.96 | No clear effect |
| use freeform optimized lens on KITTI object detection | Frustum PointNet · KITTI pedestrian 3D AP Moderate | 3D AP: 7.10 -> 11.74 | Improved |
| use freeform optimized lens on KITTI object detection | Frustum PointNet · KITTI cyclist 3D AP Moderate | 3D AP: 2.36 -> 4.89 | Improved |
| use freeform optimized lens for car 3D detection | Frustum PointNet · KITTI car 3D detection IoU 0.7 (val) | AP Moderate: 13.82 -> 17.07 | Improved |
| use freeform optimized lens on cyclist 2D detection | Frustum PointNet · KITTI cyclist 2D AP | 2D AP: 89.77 -> 88.11 | Hurt |
Closest papers by meaning
- Learning Depth from Focus in the Wild (ECCV 2022 · similarity 0.90)
- Single Image Depth Estimation Trained via Depth From Defocus Cues (CVPR 2019 · similarity 0.89)
- Blurry-Edges: Photon-Limited Depth Estimation from Defocused Boundaries (CVPR 2025 · similarity 0.88)
- MonoCD: Monocular 3D Object Detection with Complementary Depths (CVPR 2024 · similarity 0.88)
- Depth Pro: Sharp Monocular Metric Depth in Less Than a Second (ICLR 2025 · similarity 0.87)
- Repurposing Marigold for Zero-Shot Metric Depth Estimation via Defocus Blur Cues (NeurIPS 2025 · similarity 0.87)
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.
