Research map / Detection and segmentation / 3D detection and driving
Depth From Camera Motion and Object Detection
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 camera-motion and object-detection perturbations during training (ablation) | DBox · ODMD Robot Set | mean percent error: 12.5 -> 11.2 | Helped |
| Add camera-motion and object-detection perturbations during training (ablation) | DBox · ODMD All Sets | mean percent error: 5.8 -> 4.5 | Helped |
| Use relative-depth loss instead of absolute-depth loss (ablation) | DBox · ODMD Robot Set | mean percent error: 13.3 -> 11.2 | Helped |
| Use relative-depth loss instead of absolute-depth loss (ablation) | DBox · ODMS Driving | mean percent error: 53.1 -> 24.8 | Helped |
| Use only z-axis camera motion instead of full x,y,z motion (ablation) | DBox_p · ODMD All Sets | mean percent error: 4.5 -> 15.6 | Hurt |
| Train for 1 × 10^6 rather than 1 × 10^7 iterations (ablation) | DBox_p · ODMD All Sets | mean percent error: 4.5 -> 4.6 | No clear effect |
| Train for 1 × 10^5 rather than 1 × 10^7 iterations (ablation) | DBox_p · ODMD All Sets | mean percent error: 4.5 -> 5.0 | Hurt |
| Use full x,y,z motion rather than only z-axis motion (ablation) | DBox_p^z · ODMS Robot | mean percent error: 22.0 -> 11.5 | Helped |
| Train using perturbations instead of clean data (ablation) | DBox^z · ODMS Robot | mean percent error: 39.3 -> 11.5 | Helped |
| Use optical expansion over all observations rather than a two-observation motion-parallax solution (ablation) | analytical depth estimation · ODMD Robot Set | mean percent error: 65.6 -> 21.2 | Helped |
| Train with relative-depth loss and perturbations for the target motion profile (ablation) | DBox · ODMS All Sets | mean percent error: 29.3 -> 17.1 | Helped |
| Use a box-based detector input rather than segmentation input | DBox_p^z · ODMS All Sets | mean percent error: 17.8 -> 17.1 | Helped |
Closest papers by meaning
- Depth Prediction without the Sensors: Leveraging Structure for Unsupervised Learning from Monocular Videos (AAAI 2019 · similarity 0.88)
- Recurrent Neural Network for (Un-)Supervised Learning of Monocular Video Visual Odometry and Depth (CVPR 2019 · similarity 0.88)
- RM-Depth: Unsupervised Learning of Recurrent Monocular Depth in Dynamic Scenes (CVPR 2022 · similarity 0.87)
- Vision-Language Embodiment for Monocular Depth Estimation (CVPR 2025 · similarity 0.86)
- Deep Tracking: Seeing Beyond Seeing Using Recurrent Neural Networks (AAAI 2016 · similarity 0.86)
- Depth2Pose: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depth (NeurIPS 2026 · similarity 0.84)
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.