Research map / 3D vision / Depth and stereo
Sparse Auxiliary Networks for Unified Monocular Depth Prediction and Completion
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 SAN to BTS for RGB depth prediction | BTS · KITTI Eigen test | Abs.Rel | Helped |
| Add SAN to PackNet for RGB depth prediction | PackNet · KITTI Eigen test | Abs.Rel | Helped |
| Use sparse depth input with PackNet-SAN | PackNet-SAN · KITTI Eigen test | Abs.Rel | Helped |
| Add SAN to PackNet for RGB depth prediction | PackNet · DDAD camera 01 | SILog: 18.499 -> 17.767 | Helped |
| Add SAN to PackNet for RGB depth prediction | PackNet · NYUv2 test, depth prediction | AbsRel: 0.110 -> 0.106 | Helped |
| Use PackNet-SAN completion with 200 sparse samples | PackNet-SAN · NYUv2 test, completion, 200 samples | RMSE: 0.142 -> 0.155 | Hurt |
| Use PackNet-SAN completion with 500 sparse samples | PackNet-SAN · NYUv2 test, completion, 500 samples | RMSE: 0.101 -> 0.120 | Hurt |
| Use SAN-predicted depth as pseudo-LiDAR | PatchNet · KITTI3D validation, Car | AP3D@easy: 34.8/35.1 -> 35.5/35.7 | Helped |
| Use fewer branches in Sparse Residual Blocks (ablation) | PackNet-SAN · KITTI Eigen test, RGB | Abs.Rel: 0.052 -> 0.057 | Hurt |
| Unfreeze the prediction encoder during stage two (ablation) | PackNet-SAN · KITTI Eigen test, RGB+D | Abs.Rel: 0.015 -> 0.021 | Hurt |
| Freeze the prediction decoder during stage two (ablation) | PackNet-SAN · KITTI Eigen test, RGB+D | Abs.Rel: 0.015 -> 0.024 | Hurt |
| Remove shared skip-connection Wi and Bi parameters (ablation) | PackNet-SAN · KITTI Eigen test, RGB | SILog: 7.618 -> 8.324 | Hurt |
Closest papers by meaning
- A Simple yet Universal Framework for Depth Completion (NeurIPS 2024 · similarity 0.88)
- Semi-Supervised Deep Learning for Monocular Depth Map Prediction (CVPR 2017 · similarity 0.88)
- Dense Depth Posterior (DDP) From Single Image and Sparse Range (CVPR 2019 · similarity 0.88)
- Neural RGB(r)D Sensing: Depth and Uncertainty From a Video Camera (CVPR 2019 · similarity 0.87)
- Deep RGB-D Canonical Correlation Analysis For Sparse Depth Completion (NeurIPS 2019 · similarity 0.87)
- Zero-shot Depth Completion via Test-time Alignment with Affine-invariant Depth Prior (AAAI 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.