Hi there!
This is an effort to create a multi-elevation spatial digital twin of the JHU Homewood campus.
We collected visual data across the campus using a combination of drones and 360 cameras, and
deploy our image-based 3D reconstruction algorithms to reconstruct the campus. More details are
available in our paper
ULTRA-360: Unconstrained Dataset for Large-scale Temporal 3D Reconstruction across Altitudes and Omnidirectional Views
.
We have open-sourced all the collected data at
Hugging Face
.
In this interface, there are three levels of reconstruction: campus reconstruction, aerial
reconstructions of individual buildings, and ground reconstructions of individual buildings. You can
access the assets by selecting specific assets in the Scenes menu, or by clicking on
a building name, which will bring you to the aerial view of the individual building.
You can go down to the ground level from the aerial views by clicking the down button
in controls. You can also access different aerial reconstructions in weather/time by clicking the
weather icon. On the ground level, you can go up to the sky level by clicking the
up button. You can also visit nearby buildings by clicking the Nearby
button in the control menu. To simplify the viewing experience, we provide a pre-set moving trajectory.
We want to acknowledge the contribution of many students at AIEM, including Xijun Liu,
Zhaoliang Zhang, Yuxiang Guo, Yifan Zhou,
Nengyu Wang, Deming Li, and Yutao Tang, who helped
collect, calibrate, and reconstruct from 100k+ video frames over the span of two years. We want
to thank UCSD, Kaiwen Jiang and Prof. Ravi Ramamoorthi, for contributing
to the 3D reconstruction pipeline, and Mayachitra, Chandrakanth Gudavalli,
Manhar Mohammed, Prof. Shivkumar Chandrasekaran, and
Prof. B. S. Manjunath, for contributing to the image calibration pipeline. This effort
is made possible by the AAVISS project at JHU, led by Prof. Rama Chellappa,
Vishal Patel, Cheng Peng and Dr. Abhay Yadav and supported by the
IARPA WRIVA program
.
We hope you will enjoy this environment!
Cheers,
Cheng Peng