Research map / 3D vision / Depth and stereo
Multi-Object Discovery by Low-Dimensional Object Motion
Sadra Safadoust, Fatma Güney
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 depth-based translation basis to rotation-only model (ablation) | Our method · MOVi-E | FG-ARI: 73.03 -> 78.33 | Improved |
| add depth-based translation basis to rotation-only model (ablation) | Our method · MOVi-D | FG-ARI: 70.57 -> 76.41 | Improved |
| add depth-based translation basis to rotation-only model (ablation) | Our method · MOVi-A | FG-ARI: 90.43 -> 56.09 | Hurt |
| add rotation basis to translation-only model (ablation) | Our method · MOVi-C | FG-ARI: 69.69 -> 73.80 | Improved |
| model pixel-wise depth with low-dimensional flow bases instead of predicting motion distributions | single-image multi-object segmentation · MOVi-E | FG-ARI: 62.62 -> 78.33 | Improved |
| apply post-processing (ablation) | Our method · MOVi-E | FG-ARI: 78.33 -> 80.83 | Improved |
| merge predicted regions by spectral clustering to ground truth number of regions (Ours-M) | Our method · DAVIS-2017 | J&F: 55.3 -> 59.2 | Improved |
| use known camera intrinsics (6 bases instead of 8) | Our depth network · KITTI Eigen split | Abs Rel: 0.107 -> 0.084 | Improved |
| train depth jointly with rotation, vs translation-only (ablation) | depth network · MOVi-D | Abs Rel: 0.357 -> 0.544 | Hurt |
Closest papers by meaning
- Unsupervised Multi-Object Segmentation by Predicting Probable Motion Patterns (NeurIPS 2022 · similarity 0.91)
- Discovering Objects That Can Move (CVPR 2022 · similarity 0.90)
- Competitive Collaboration: Joint Unsupervised Learning of Depth, Camera Motion, Optical Flow and Motion Segmentation (CVPR 2019 · similarity 0.89)
- Learning Motion Patterns in Videos (CVPR 2017 · similarity 0.89)
- Video Segmentation via Object Flow (CVPR 2016 · similarity 0.89)
- Every Pixel Counts ++: Joint Learning of Geometry and Motion with 3D Holistic Understanding (TPAMI 2019 · similarity 0.89)
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.
