Introducing 4D Primitive-Mâché (4DPM), a new method for replayable 4D reconstruction from monocular videos.
We split dynamic scenes into 3D primitives and recover their motion. 4DPM can infer object positions even after they leave view.
Joint work with @marwan_ptr@AjdDavison
Introducing 4D Primitive-Mâché (4DPM), a new method for replayable 4D reconstruction from monocular videos.
We split dynamic scenes into 3D primitives and recover their motion. 4DPM can infer object positions even after they leave view.
Joint work with @marwan_ptr@AjdDavison
Introducing 4D Primitive-Mâché (4DPM), a new method for replayable 4D reconstruction from monocular videos.
We split dynamic scenes into 3D primitives and recover their motion. 4DPM can infer object positions even after they leave view.
Joint work with @marwan_ptr@AjdDavison
How can we run reconstruction models like π³ and Depth Anything 3 in real-time?
We present KV-Tracker, a training-free approach, for real-time tracking of scenes and objects. Achieving up to 30 FPS!
With @alzugarayign, @makezur, @XinKong_IC and @AjdDavison
@sha0nk each pixel is only assigned to (at most) one primitive; all segments are also eroded around the edges to compensate for feed-forward reconstruction artifacts around object boundaries
Introducing 4D Primitive-Mâché (4DPM), a new method for replayable 4D reconstruction from monocular videos.
We split dynamic scenes into 3D primitives and recover their motion. 4DPM can infer object positions even after they leave view.
Joint work with @marwan_ptr@AjdDavison
Our method cuts 3D primitives out of the outputs of a feed-forward reconstruction model (π³) and then glues them across time.
Each primitive’s motion is represented compactly as a single SE(3) pose, inferred from estimated 2D correspondences via an optimisation pipeline.
4DPM handles multiple moving entities with distinct trajectories.
The result is a complete, replayable 4D reconstruction of all observed scene elements at every timestamp.
Introducing MASt3R-SLAM, the first real-time monocular dense SLAM with MASt3R as a foundation.
Easy to use like DUSt3R/MASt3R, from an uncalibrated RGB video it recovers accurate, globally consistent poses & a dense map.
With @eric_dexheimer*, @AjdDavison (*Equal Contribution)
SuperPrimitives will be presented at #CVPR next week (Wednesday), along with a 𝗿𝗲𝗮𝗹-𝘁𝗶𝗺𝗲 𝗱𝗲𝗺𝗼 on Friday!
Our new representation enables dense monocular 3D reconstruction in real-time. No poses required!
Project page: https://t.co/l360soWiAz
Curious how DSINE surface normals can be converted into full 3D geometry?
Visit our demo of 𝗦𝘂𝗽𝗲𝗿𝗣𝗿𝗶𝗺𝗶𝘁𝗶𝘃𝗲-based real-time 3D reconstruction tomorrow (Friday)! A demo is worth a thousand pictures 🙂
Demo: 10:30 to 18:45, Arch CDE #9
#CVPR#CVPR2024
SuperPrimitives will be presented at #CVPR next week (Wednesday), along with a 𝗿𝗲𝗮𝗹-𝘁𝗶𝗺𝗲 𝗱𝗲𝗺𝗼 on Friday!
Our new representation enables dense monocular 3D reconstruction in real-time. No poses required!
Project page: https://t.co/l360soWiAz
SuperPrimitives will be presented at #CVPR next week (Wednesday), along with a 𝗿𝗲𝗮𝗹-𝘁𝗶𝗺𝗲 𝗱𝗲𝗺𝗼 on Friday!
Our new representation enables dense monocular 3D reconstruction in real-time. No poses required!
Project page: https://t.co/l360soWiAz