Research map / 3D vision / Depth and stereo
Semi-Supervised Deep Learning for Monocular Depth Map Prediction
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 stereo image-alignment supervision to supervised laser-depth training | ResNet-50 encoder-decoder depth predictor · KITTI Raw Eigen test split, 0–80 m cap | RMSE: 4.815 -> 4.621 | Helped |
| Add stereo image-alignment supervision to supervised laser-depth training | ResNet-50 encoder-decoder depth predictor · KITTI Raw Eigen test split, 0–80 m cap | δ < 1.25: 0.845 -> 0.862 | Helped |
| Use supervised laser-depth training instead of unsupervised training alone | ResNet-50 encoder-decoder depth predictor · KITTI Raw Eigen test split, 0–80 m cap | RMSE: 8.700 -> 4.621 | Helped |
| Add stereo image-alignment supervision to supervised laser-depth training | ResNet-50 encoder-decoder depth predictor · KITTI Raw Eigen test split, uncropped and uncapped; GT minimum 5 m | RMSE: 4.862 -> 4.627 | Helped |
| Use the unsupervised loss at all valid pixels rather than only pixels without ground truth (ablation) | Our full approach · KITTI Raw Eigen test split, uncropped and uncapped; GT minimum 5 m | RMSE: 4.679 -> 4.627 | Helped |
| Use only 50% of laser points (ablation) | Our full approach · KITTI Raw Eigen test split, uncropped and uncapped; GT minimum 5 m | RMSE: 4.627 -> 4.808 | Hurt |
| Use only 1% of laser points (ablation) | Our full approach · KITTI Raw Eigen test split, uncropped and uncapped; GT minimum 5 m | RMSE: 4.627 -> 4.892 | Hurt |
| Add long skip connections and Gaussian smoothing (ablation) | Our full approach · KITTI Raw Eigen test split, uncropped and uncapped; GT minimum 5 m | RMSE: 4.798 -> 4.627 | Helped |
| Add long skip connections (ablation) | Our full approach · KITTI Raw Eigen test split, uncropped and uncapped; GT minimum 5 m | RMSE: 4.762 -> 4.627 | Helped |
| Add Gaussian smoothing to the unsupervised loss (ablation) | Our full approach · KITTI Raw Eigen test split, uncropped and uncapped; GT minimum 5 m | RMSE: 4.752 -> 4.627 | Helped |
| Replace L2 with BerHu in the supervised loss (ablation) | Our full approach · KITTI Raw Eigen test split, uncropped and uncapped; GT minimum 5 m | RMSE: 4.659 -> 4.627 | Helped |
Closest papers by meaning
- Dense Depth Posterior (DDP) From Single Image and Sparse Range (CVPR 2019 · similarity 0.89)
- Mind The Edge: Refining Depth Edges in Sparsely-Supervised Monocular Depth Estimation (CVPR 2024 · similarity 0.89)
- Unsupervised Monocular Depth Estimation With Left-Right Consistency (CVPR 2017 · similarity 0.89)
- CNN-SLAM: Real-Time Dense Monocular SLAM With Learned Depth Prediction (CVPR 2017 · similarity 0.88)
- From-Ground-To-Objects: Coarse-to-Fine Self-supervised Monocular Depth Estimation of Dynamic Objects with Ground Contact Prior (CVPR 2024 · similarity 0.88)
- Distilling Monocular Foundation Model for Fine-grained Depth Completion (CVPR 2025 · 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.