CV4DT

The Computer Vision for Digital Twins (CV4DT) is a research group based at the University of Cambridge and initiated by Olaf Wysocki. The CV4DT centres on developing methods and datasets for pushing the boundaries of 3D computer vision and robotics for accurate transfer of reality into the digital world to enable simulation before any action is taken.

2026-09 — 3x papers accepted for the NeurIPS 2026!
2026-06 — 3x papers accepted for the ECCV 2026!
2026-09 — We organise the TwinWorld workshops at the ECCV 2026!
2026-06 — Our GS4City paper goes to CVPR 2026!
2026-05 — Our collaborative EnerGS paper goes to ICML 2026!
2025-10 — Our ActionReasoning paper goes to ICRA 2026!
2026-01 — We'll be co-organsing 2x CVPR 2026 workshops: PBVS and DriveX!

Older news

2025-12 — 3x 3D Computer Vision and Robotics projects together with TUM (more)
2025-11 — Brian gives a keynote about multimodality at BMVC 2025!
2025-10 — Our TrueCity paper goes to 3DV 2026!
2025-08 — The CV4DT is born!

👋 Welcome to the CV4DT world

Check out our research focus…

3D Semantic Understanding

3D scene understanding with imbalanced data

3D Semantic Object Reconstruction

Reconstructing high-detail semantic 3D models

3D Models as Novel Modality

High-detail structured 3D models as new sensor signal

Digital Twins for World Models

3D-vision-informed world models for robotics

Latest Papers

Rui Song, Tianhui Cai, Markus Gross, Yun Zhang, Walter Zimmer, Zhiyu Huang, Olaf Wysocki, Jiaqi Ma (2026). EnerGS: Energy-Based Gaussian Splatting with Partial Geometric Priors.
ICML ‘26.

Guangming Wang, Qizhen Ying, Yixiong Jing, Olaf Wysocki, Brian Sheil (2026). ActionReasoning: Robot Action Reasoning in 3D Space with LLM for Robotic Brick Stacking.
ICRA ‘26.

Duc Nguyen, Yan-Ling Lai, Qilin Zhang, Prabin Gyawali, Benedikt Schwab, Olaf Wysocki, Thomas H Kolbe (2025). TrueCity: Real and Simulated Urban Data for Cross-Domain 3D Scene Understanding.
3DV ‘26.

Cite arXiv PDF Project page

Shuhao Kang, Martin Y Liao, Yan Xia, Olaf Wysocki, Boris Jutzi, Daniel Cremers (2025). OPAL: Visibility-aware Lidar-to-OpenStreetMap Place Recognition via Adaptive Radial Fusion.
CoRL ‘25.

Wenzhao Tang, Weihang Li, Xiucheng Liang, Olaf Wysocki, Filip Biljecki, Christoph Holst, Boris Jutzi (2025). Texture2LoD3: Enabling LoD3 Building Reconstruction With Panoramic Images.
CVPR ‘25.

Cite arXiv PDF Project page

See all publications