Research map / 3D vision / Depth and stereo
Dynamic Point Maps: A Versatile Representation for Dynamic 3D Reconstruction
Edgar Sucar, Zihang Lai, Eldar Insafutdinov, Andrea Vedaldi
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 |
|---|---|---|---|
| replace MonST3R with DPM for two-view depth (Kubric) | MonST3R · Two-view depth, Kubric | Abs Rel: 0.166 -> 0.078 | Improved |
| replace MonST3R with DPM for two-view depth (Bonn) | MonST3R · Two-view depth, Bonn | Abs Rel: 0.071 -> 0.082 | Hurt |
| replace MonST3R with DPM for two-view depth (Sintel) | MonST3R · Two-view depth, Sintel | Abs Rel: 0.347 -> 0.321 | Improved |
| replace MonST3R with DPM for video depth (KITTI) | MonST3R · Video depth, KITTI | Abs Rel: 0.104 -> 0.140 | Hurt |
| replace MonST3R with DPM for video depth (KITTI crop) | MonST3R · Video depth, KITTI (crop) | Abs Rel: 0.111 -> 0.097 | Improved |
| remove optical flow loss from bundle adjustment when using DPM | DPM video depth · Video depth | Abs Rel | No clear effect |
| predict cross-time point maps with DPM instead of MonST3R+RAFT | MonST3R+RAFT · Kubric-F dynamic reconstruction | L_rel P1(t2): 0.394 -> 0.049 | Improved |
| predict cross-time point maps with DPM (Kubric-G) | MonST3R+RAFT · Kubric-G dynamic reconstruction | L_rel P1(t2): 0.346 -> 0.079 | Improved |
| use DPM for object pose tracking | MonST3R+RAFT · Kubric-F object pose | RPE rot: 56.1 -> 33.7 | Improved |
| use DPM for scene flow instead of MonST3R | MonST3R · Scene flow, Kubric-G forward | 3D EPE: 0.334 -> 0.104 | Improved |
| use RGB-only DPM instead of RGBD RAFT-3D for scene flow (Kubric-F) | RAFT-3D · Scene flow forward, Kubric-F | 3D EPE: 0.051 -> 0.081 | Hurt |
| use RGB-only DPM instead of RGBD RAFT-3D for scene flow (Waymo) | RAFT-3D · Scene flow forward, Waymo | 3D EPE: 0.150 -> 0.051 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- MonST3R — MonST3R: A Simple Approach for Estimating Geometry in the Presence of Motion (ICLR 2025) · better on 9 of 12 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- MonST3R: A Simple Approach for Estimating Geometry in the Presence of Motion (ICLR 2025 · similarity 0.91)
- Enhancing 3D Reconstruction for Dynamic Scenes (NeurIPS 2025 · similarity 0.91)
- Stereo4D: Learning How Things Move in 3D from Internet Stereo Videos (CVPR 2025 · similarity 0.90)
- DualPM: Dual Posed-Canonical Point Maps for 3D Shape and Pose Reconstruction (CVPR 2025 · similarity 0.89)
- DUSt3R: Geometric 3D Vision Made Easy (CVPR 2024 · similarity 0.89)
- 4D3R: Motion-Aware Neural Reconstruction and Rendering of Dynamic Scenes from Monocular Videos (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.
