Ilya Fradlin    Christian Schmidt    Jens Piekenbrinck
Karim Knaebel    Gonzalo Martin Garcia    Bastian Leibe

RWTH Aachen University

• 2026-10-01: Our paper is available on arXiv.
• Coming soon: code and weights. Watch this repository for release updates.

ARROW is a feed-forward model that unifies 3D reconstruction and 3D point tracking from arbitrary image sets. It combines the flexible point-query decoding of D4RT with support for arbitrary observations across viewpoints and capture times. This enables reconstruction and tracking from moving-camera videos, multiple video streams, and unordered photo collections.

ARROW achieves state-of-the-art 3D tracking on WorldTrack and TAPVid-3D and outperforms dedicated multi-view trackers on an adapted RGB-only MVTracker benchmark, remaining competitive on 3D reconstruction tasks.

Visit the project page for interactive examples and results.

@article{fradlin2026arrow,
    title         = {{ARROW}: Arbitrary Reconstruction and Tracking of 4D Observations in the Wild},
    author        = {Fradlin, Ilya and Schmidt, Christian and Piekenbrinck, Jens and Knaebel, Karim and Martin Garcia, Gonzalo and Leibe, Bastian},
    year          = 2026,
    journal       = {arXiv preprint arXiv:2610.01314},
}

This project is licensed under the MIT License.