Research map / 3D vision / Depth and stereo
A Two-Streamed Network for Estimating Fine-Scaled Depth Maps From Single RGB Images
Jun Li, Reinhard Klein, Angela Yao
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 set image loss regularizer to depth-only VGG-16 (ablation) | depth stream, VGG-16 · NYU Depth v2 (654 test images) | rms: 0.640 -> 0.617 | Improved |
| add set image loss regularizer to depth-only VGG-16 (ablation) | depth stream, VGG-16 · NYU Depth v2 (654 test images) | rel error: 0.161 -> 0.153 | Improved |
| add set image loss regularizer to depth-only VGG-16 (ablation) | depth stream, VGG-16 · NYU Depth v2 (654 test images) | delta<1.25 accuracy: 0.765 -> 0.786 | Improved |
| add bilateral filtering to set-loss depth output (ablation) | L_set depth only, VGG-16 · NYU Depth v2 (654 test images) | rms: 0.617 -> 0.621 | No clear effect |
| add depth gradient stream fused by optimization (ablation) | L_set depth only, VGG-16 · NYU Depth v2 (654 test images) | rms: 0.617 -> 0.611 | Improved |
| fuse depth and gradients end-to-end instead of optimization (ablation) | depth + gradients, VGG-16 · NYU Depth v2 (654 test images) | rms: 0.611 -> 0.615 | No clear effect |
| use VGG-19 base instead of VGG-16 (ablation) | depth + gradients, optimization · NYU Depth v2 (654 test images) | rel error: 0.152 -> 0.146 | Improved |
| use ResNet-50 base instead of VGG-16 (ablation) | depth + gradients, optimization · NYU Depth v2 (654 test images) | rms: 0.611 -> 0.635 | Hurt |
| use ResNet-50 base instead of VGG-16 (ablation) | depth + gradients, optimization · NYU Depth v2 (654 test images) | rel error: 0.152 -> 0.143 | Improved |
| add rotation and translation to the image set (ablation) | set loss · NYU Depth v2 (654 test images) | depth accuracy | No clear effect |
| increase set loss weight lambda (ablation) | set loss · NYU Depth v2 (654 test images) | depth accuracy | No clear effect |
| increase gradient weight omega in optimization fusion (ablation) | optimization fusion · NYU Depth v2 (654 test images) | depth accuracy | Hurt |
Closest papers by meaning
- Depth-aware CNN for RGB-D Segmentation (ECCV 2018 · similarity 0.89)
- Unsupervised CNN for Single View Depth Estimation: Geometry to the Rescue (ECCV 2016 · similarity 0.88)
- Learning Joint 2D-3D Representations for Depth Completion (ICCV 2019 · similarity 0.88)
- Single Image Depth Estimation Trained via Depth From Defocus Cues (CVPR 2019 · similarity 0.88)
- Deep Depth Completion of a Single RGB-D Image (CVPR 2018 · similarity 0.88)
- CostDCNet: Cost Volume Based Depth Completion for a Single RGB-D Image (ECCV 2022 · similarity 0.88)
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.
