Research map / 3D vision / Depth and stereo
Feature-Level Collaboration: Joint Unsupervised Learning of Optical Flow, Stereo Depth and Camera Motion
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 |
|---|---|---|---|
| Share image features and jointly decode flow and disparity (ablation) | Flow estimation · KITTI 2012 and KITTI 2015 train | EPE-noc: 1.04 to 0.94; 3.38 to 2.95 | Helped |
| Use complementary stereo/flow cost volumes (ablation) | Flow estimation · KITTI 2012 and KITTI 2015 train | EPE-occ: 9.99 to 7.40; 16.53 to 12.69 | Helped |
| Add pose-based flow consistency constraint (ablation) | Feature Fusion · KITTI 2012 train | EPE-all: 1.93 to 1.35 | Helped |
| Add pose-based flow consistency constraint (ablation) | Cost Volume Fusion · KITTI 2012 train | EPE-all: 1.87 to 1.25 | Helped |
| Add pose-based flow consistency constraint (ablation) | Feature Fusion · KITTI 2015 train | EPE-all: 5.51 to 3.08 | Helped |
| Add pose-based flow consistency constraint (ablation) | Cost Volume Fusion · KITTI 2015 train | EPE-all: 5.06 to 2.35 | Helped |
| Use complementary cost volume (ablation) | Feature Fusion + Pose · KITTI 2015 static regions | EPE-static: 2.56 to 1.84 | Helped |
| Apply cost-volume complement using camera pose in moving regions (ablation) | Optical flow estimation · KITTI 2015 moving regions | EPE-moving: 4.55 to 10.05 | Hurt |
| Add moving-object flow handling module (ablation) | Cost Volume Fusion + Pose(static) · KITTI 2015 moving regions | EPE-moving: 10.05 to 3.30 | Helped |
| Estimate camera pose from shared flow/depth features (ablation) | Camera motion estimation · KITTI VO Seq. 09 | terr%: 13.98 to 5.77 | Helped |
| Add full cost-volume complement and flow refinement to the motion module (ablation) | Camera motion estimation · KITTI VO Seq. 09 | terr%: 5.77 to 2.02 | Helped |
Closest papers by meaning
- Bridging Stereo Matching and Optical Flow via Spatiotemporal Correspondence (CVPR 2019 · similarity 0.93)
- GeoNet: Unsupervised Learning of Dense Depth, Optical Flow and Camera Pose (CVPR 2018 · similarity 0.92)
- UnOS: Unified Unsupervised Optical-Flow and Stereo-Depth Estimation by Watching Videos (CVPR 2019 · similarity 0.91)
- Competitive Collaboration: Joint Unsupervised Learning of Depth, Camera Motion, Optical Flow and Motion Segmentation (CVPR 2019 · similarity 0.90)
- Flow-NeRF: Joint Learning of Geometry, Poses, and Dense Flow within Unified Neural Representations (CVPR 2025 · similarity 0.89)
- Depth2Pose: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depth (NeurIPS 2026 · 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.