Research map / Label-efficient and robust learning / Domain adaptation and generalization
Real-Time Monocular Depth Estimation Using Synthetic Data With Domain Adaptation via Image Style Transfer
Read the paperCheck 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 |
|---|---|---|---|
| train on synthetic paired RGB and depth data and adapt real images to the synthetic domain with cycleGAN | Our monocular depth estimator · KITTI [18] split | Abs. Rel.: 0.124 -> 0.110 | Helped |
| train on synthetic paired RGB and depth data and adapt real images to the synthetic domain with cycleGAN | Our monocular depth estimator · KITTI [18] split | RMSE: 5.311 -> 4.726 | Helped |
| remove domain adaptation (ablation) | Our approach · KITTI | Abs. Rel.: 0.101 -> 0.498 | Hurt |
| remove domain adaptation (ablation) | Our approach · KITTI | RMSE: 5.308 -> 9.382 | Hurt |
| replace cycleGAN domain adaptation with Johnson et al. style transfer (ablation) | Our approach · KITTI | Abs. Rel.: 0.101 -> 0.154 | Hurt |
| replace cycleGAN domain adaptation with Johnson et al. style transfer (ablation) | Our approach · KITTI | RMSE: 5.308 -> 6.470 | Hurt |
| use cycleGAN domain adaptation rather than no domain adaptation (ablation) | Our approach · KITTI | Sq. Rel.: 6.533 -> 1.048 | Helped |
| use cycleGAN domain adaptation rather than no domain adaptation (ablation) | Our approach · KITTI | RMSE log: 0.609 -> 0.184 | Helped |
| combine reconstruction loss with adversarial loss instead of using reconstruction loss alone (ablation) | Synthetic-data depth estimator · Synthetic test data | depth-map sharpness and accuracy | Helped |
| use reconstruction loss alongside adversarial loss instead of adversarial loss alone (ablation) | Synthetic-data depth estimator · Synthetic test data | depth-map artefacts | Helped |
| apply the model to Make3D without training on that dataset | Our approach · Make3D | Abs. Rel.: 0.189 -> 0.423 | Hurt |
| apply the model to Make3D without training on that dataset | Our approach · Make3D | RMSE: 5.285 -> 9.002 | Hurt |
Closest papers by meaning
- DepthFM: Fast Generative Monocular Depth Estimation with Flow Matching (AAAI 2025 · similarity 0.88)
- Synthetic-to-Real Self-supervised Robust Depth Estimation via Learning with Motion and Structure Priors (CVPR 2025 · similarity 0.86)
- DCDepth: Progressive Monocular Depth Estimation in Discrete Cosine Domain (NeurIPS 2024 · similarity 0.85)
- Repurposing Diffusion-Based Image Generators for Monocular Depth Estimation (CVPR 2024 · similarity 0.85)
- DME: Unveiling the Bias for Better Generalized Monocular Depth Estimation (AAAI 2024 · similarity 0.85)
- Scalable Autoregressive Monocular Depth Estimation (CVPR 2025 · similarity 0.85)
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.