Computer Vision: A Technical Deep Dive into the Method and Its Applications in Urban Space and History Research

What mechanistic processes occur when we say that a computer "sees"? What does it mean when we "train" artificial neural networks to recognize images? How can these technologies help in processing visual data, and what possibilities for interacting with urban space and history might they open up?

This lecture is an attempt to combine technical information about modern computer vision methods with several examples of how the technology is applied in the analysis and presentation of visual data. In particular, we will focus on deep neural networks, methods for training them, and the so-called "structure from motion" (SfM) technique, which converts a set of 2D images into a 3D structure. The lecture will emphasize both a technical overview with a detailed dive into how neural networks work and a review of several case studies using computer vision and visual city data.

The event will take place online. To receive an invitation to participate, please register via this link.

The presentation will be in Ukrainian with simultaneous translation into English.

The event is part of the series of seminars and workshops “Urban History Digital Infrastructures: Sharing Expertise and Networking Cities,” organized by the Center for Urban History with the support of the Center for Governance and Culture in Europe at the University of St. Gallen.