Cornell CS Associate Professor Noah Snavely recently received the Helmholtz Prize at the International Conference on Computer Vision (ICCV) 2019 for a paper he published a decade ago: “Building Rome in a Day.” The prize celebrates, in part, papers that have stood the test of time over the past ten years. Snavely notes: “This award is particularly meaningful to me, because this paper was the first one I worked on after joining Cornell.” He recalls his arrival on campus: “I remember working on the project in my then-empty office in Upson Hall (I had newly moved in and had just a chair and a desk).”
Snavely, who now sits at the Cornell Tech campus in New York City, summarizes the key contribution of the paper: “we show how to take huge collections of unorganized tourist photos spanning an entire city—say, hundreds of thousands of public images of Rome on a site like Flickr—and turn them into 3D models using computer vision methods run on a large cluster of machines.” Details of the project are available at this website, which includes images and videos, but Snavely adds quickly that that research “also spurred an effort at Cornell called BigSFM on reconstructing the world from Internet photos.”
Further details on the paper:
Entering the search term Rome on Flickr returns more than two million photographs. This collection represents an increasingly complete photographic record of the city, capturing every popular site, facade, interior, fountain, sculpture, painting, cafe, and so forth. It also offers us an unprecedented opportunity to richly capture, explore and study the three-dimensional shape of the city.
In this project, we consider the problem of reconstructing entire cities from images harvested from the web. Our aim is to build a parallel distributed system that downloads all the images associated with a city, say Rome, from Flickr.com. After downloading, it matches these images to find common points and uses this information to compute the three-dimensional structure of the city and the pose of the cameras that captured these images. All this to be done in a day.
This poses new challenges for every stage of the 3D reconstruction pipeline, from image matching to large scale optimization. The key contributions of our work is a new, parallel distributed matching system that can match massive collections of images very quickly and a new bundle adjust software that can solve extremely large non-linear least squares problems that are encountered in three dimensional reconstruction problems.