DrivingScene: A Multi-Task Online Feed-Forward 3D Gaussian Splatting Method for Dynamic Driving Scenes
Qirui Hou, Wenzhang Sun, Chang Zeng, Chunfeng Wang, Hao Li, Jianxun Cui · 2025
Read DrivingScene: A Multi-Task Online Feed-Forward 3D Gaussian Splatting Method for Dynamic Driving Scenes by Qirui Hou, Wenzhang Sun, Chang Zeng, Chunfeng Wang, Hao Li, Jianxun Cui (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2510-24734.txt.
Description
Qirui Hou, Wenzhang Sun, Chang Zeng, Chunfeng Wang, Hao Li, and Jianxun Cui rebuild moving street scenes in three dimensions as a drive goes on. Past tools faced a trade off between sharp views and speed, with few side shots to work from. Their tool, DrivingScene, works from just two frames of wide view shots at a time.
A small flow net in the tool gauges how moving things shift, on top of a learned base scene. Training moves from rough to fine steps to dodge the wobbles of end to end training. On a known drive set, the tool gave depth, flow, and point clouds at once and beat past tools.
Catalog record
| CREATOR | Qirui Hou, Wenzhang Sun, Chang Zeng, Chunfeng Wang, Hao Li, Jianxun Cui |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 20 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2510-24734