Research map / 3D vision / Depth and stereo
PLADE-Net: Towards Pixel-Level Accuracy for Self-Supervised Single-View Depth Estimation With Neural Positional Encoding and Distilled Matting Loss
Juan Luis Gonzalez, Munchurl Kim
Check 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 neural positional encoding (ablation) | PLADE-Net, stage 1 · KITTI improved Eigen test split, K+CS | δ¹: 0.938 -> 0.942 | Improved |
| add direct pixel-coordinate positional encoding instead of no positional encoding (ablation) | PLADE-Net, stage 1 · KITTI improved Eigen test split, K+CS | δ¹: 0.938 -> 0.939 | No clear effect |
| add neural positional encoding (ablation) | PLADE-Net, stage 2 · KITTI improved Eigen test split, K+CS | δ¹: 0.942 -> 0.949 | Improved |
| add direct pixel-coordinate positional encoding instead of no positional encoding (ablation) | PLADE-Net, stage 2 · KITTI improved Eigen test split, K+CS | δ¹: 0.942 -> 0.945 | Improved |
| add neural positional encoding rather than direct coordinate encoding (ablation) | PLADE-Net, stage 2 · KITTI improved Eigen test split, K+CS | δ¹: 0.945 -> 0.949 | Improved |
| add distilled matting Laplacian loss (ablation) | PLADE-Net with no proposed losses · KITTI improved Eigen test split, stage 2, K+CS | δ¹: 0.943 -> 0.948 | Improved |
| add deep corr-L1 loss (ablation) | PLADE-Net with no proposed losses · KITTI improved Eigen test split, stage 2, K+CS | δ¹: 0.943 -> 0.945 | Improved |
| add distilled matting Laplacian loss rather than deep corr-L1 loss (ablation) | PLADE-Net · KITTI improved Eigen test split, stage 2, K+CS | δ¹: 0.945 -> 0.948 | Improved |
| add deep corr-L1 loss alongside the distilled matting loss (ablation) | PLADE-Net with matting loss · KITTI improved Eigen test split, stage 2, K+CS | δ¹: 0.948 -> 0.948 | No clear effect |
| use K+CS training rather than K-only training (ablation) | PLADE-Net with NPE, stage 2 · KITTI improved Eigen test split | δ¹: 0.944 -> 0.949 | Improved |
Closest papers by meaning
- BRNet: Exploring Comprehensive Features for Monocular Depth Estimation (ECCV 2022 · similarity 0.91)
- Revealing the Reciprocal Relations Between Self-Supervised Stereo and Monocular Depth Estimation (ICCV 2021 · similarity 0.91)
- Self-Supervised Monocular Depth Hints (ICCV 2019 · similarity 0.90)
- Digging Into Self-Supervised Monocular Depth Estimation (ICCV 2019 · similarity 0.90)
- Reversing the cycle: self-supervised deep stereo through enhanced monocular distillation (ECCV 2020 · similarity 0.90)
- Self-Supervised Monocular Depth Estimation by Direction-aware Cumulative Convolution Network (ICCV 2023 · 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.
