Research map / 3D vision / Depth and stereo
EffiScene: Efficient Per-Pixel Rigidity Inference for Unsupervised Joint Learning of Optical Flow, Depth, Camera Pose and Motion Segmentation
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 |
|---|---|---|---|
| jointly infer rigidity from optical flow, depth, and camera pose | EffiScene · KITTI 2015 train | EPE All: 5.19 -> 4.20 | Helped |
| jointly infer rigidity from optical flow, depth, and camera pose | EffiScene · KITTI 2015 train | RMSE: 3.780 -> 3.461 | Helped |
| jointly infer rigidity from optical flow, depth, and camera pose | EffiScene · KITTI 2015 train | Mean IoU: 0.570 -> 0.615 | Helped |
| jointly infer rigidity from optical flow, depth, and camera pose | EffiScene · KITTI Odometry sequence 09 | ATE: 0.012$\pm$0.007 -> 0.011$\pm$0.006 | Helped |
| use the fused rigid and optical flows on a scene with no moving objects | EffiScene · KITTI 2012 train | EPE All: 1.64 -> 1.68 | Hurt |
| use the RfM rigidity constraint and jointly optimize FlowCNN and DepthCNN (ablation) | EffiScene · KITTI 2015 train | EPE All: 6.76 -> 4.70 | Helped |
| fuse optical flow with rigid flow using the learned rigidity map (ablation) | EffiScene · KITTI 2015 train | EPE All: 4.70 -> 4.20 | Helped |
| replace RAFT with PWC-Net as the flow backbone (ablation) | EffiScene · KITTI 2015 train | EPE All: 4.20 -> 4.92 | Hurt |
| replace the P$n$P solver with a pose-regression network (ablation) | EffiScene · KITTI Odometry sequence 09 | ATE: 0.011$\pm$0.006 -> 0.013$\pm$0.006 | Hurt |
Closest papers by meaning
- GeoNet: Unsupervised Learning of Dense Depth, Optical Flow and Camera Pose (CVPR 2018 · similarity 0.89)
- PointFlowNet: Learning Representations for Rigid Motion Estimation From Point Clouds (CVPR 2019 · similarity 0.88)
- UnSAMFlow: Unsupervised Optical Flow Guided by Segment Anything Model (CVPR 2024 · similarity 0.87)
- VoteFlow: Enforcing Local Rigidity in Self-Supervised Scene Flow (CVPR 2025 · similarity 0.87)
- E-MoFlow: Learning Egomotion and Optical Flow from Event Data via Implicit Regularization (NeurIPS 2025 · similarity 0.86)
- Flow-NeRF: Joint Learning of Geometry, Poses, and Dense Flow within Unified Neural Representations (CVPR 2025 · similarity 0.86)
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.