Research map / 3D vision / Depth and stereo
SynDeMo: Synergistic Deep Feature Alignment for Joint Learning of Depth and Ego-Motion
Behzad Bozorgtabar, Mohammad Saeed Rad, Dwarikanath Mahapatra, Jean-Philippe Thiran
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 synthetic data with task, multi-view and synergistic feature alignment losses (ablation) | SynDeMo baseline · KITTI Eigen test (train K+vK) | Abs Rel: 0.196 -> 0.116 | Improved |
| add synergistic deep feature alignment (with task loss and multi-view loss) (ablation) | SynDeMo · KITTI Eigen test (train K+vK) | Abs Rel: 0.123 -> 0.116 | Improved |
| add multi-view self-supervised loss on synthetic images (ablation) | SynDeMo · KITTI Eigen test (train K+vK) | Abs Rel: 0.126 -> 0.116 | Improved |
| add supervised task loss on labeled synthetic depth (ablation) | SynDeMo · KITTI Eigen test (train K+vK) | Abs Rel: 0.145 -> 0.116 | Improved |
| add Cityscapes to training data | SynDeMo · KITTI Eigen test | Abs Rel: 0.116 -> 0.112 | Improved |
| add synthetic geometry information to handle moving objects (ablation) | SynDeMo · Cityscapes dynamic scenes | qualitative depth | Improved |
| add full SynDeMo objectives (ablation) | Baseline · KITTI odometry Seq. 09 | ATE: 0.019 -> 0.011 | Improved |
| add full SynDeMo objectives (ablation) | Baseline · Make3D (zero-shot) | Abs Rel: 0.375 -> 0.295 | Improved |
| add feature alignment and task loss (ablation) | SynDeMo · Make3D (zero-shot) | Abs Rel: 0.311 -> 0.295 | Improved |
Closest papers by meaning
- Discriminately Treating Motion Components Evolves Joint Depth and Ego-Motion Learning (TPAMI 2026 · similarity 0.92)
- Unsupervised Learning of Depth and Ego-Motion From Monocular Video Using 3D Geometric Constraints (CVPR 2018 · similarity 0.92)
- Depth2Pose: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depth (NeurIPS 2026 · similarity 0.89)
- Mining Supervision for Dynamic Regions in Self-Supervised Monocular Depth Estimation (CVPR 2024 · similarity 0.89)
- Seurat: From Moving Points to Depth (CVPR 2025 · similarity 0.88)
- E-MoFlow: Learning Egomotion and Optical Flow from Event Data via Implicit Regularization (NeurIPS 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.
