Distant Viewing Toolkit: Software for analysing visual culture

Distant Viewing Toolkit: Software for analysing visual culture,

Lauren Tilton, Assistant Professor of Digital Humanities, Department of Rhetoric and Communication, University of Richmond and Taylor Arnold, Assistant Professor of Statistics and Linguistics, Department of Mathematics and Computer Science, University of Richmond

You can see this video (enhanced version) in fullscreen here

Abstract

This video introduces the "Distant Viewing Toolkit" (DVT), a Python package for computational analysis of visual culture. The package addresses the challenges of working with moving images through the automated extraction and visualization of metadata summarizing content (e.g. people / actors, dialogue, scenes, objects) and style (e.g. camera angle view, lighting, framing, sound). In this presentation, we highlight the main characteristics of the tools and the types of humanistic analyzes that the toolkit makes possible. The video will explain design decisions, common use cases, and how the toolkit fits into the larger landscape of computational tools in the study of visual culture.

The toolkit offers two interfaces. A low-level API provides “annotators,” algorithms that work directly with the source data, and “aggregators” that have access to information pulled from all previously executed annotators and aggregators across the entire input but cannot directly access visual data. The separation of the algorithms into these two parts allows easy and error-free code writing and faithfully reflects the theory of Distant Viewing. The second interface, a command line tool, provides a quick way to get started with the tools. The development of DVT follows best practices for developing open source software. The project has an open source license (GPL-2), uses the “Covenant Code of Conduct” contributor (v1.4) and provides user models for submitting issues and pull requests.

Biographies
Taylor Arnold is an Assistant Professor at the University of Richmond, Virginia, in the Linguistics Program, Department of Mathematics and Computer Science. In 2013, he received a doctorate in statistics from Yale University. Prior to his current role, he was a senior scientist with AT&T Research Laboratories in New York. In 2019-2020, he was a resident researcher at the Collegium de Lyon. He studies large cultural datasets in the framework of research projects in the humanities and social sciences, both new and existing. As such, he specializes in the application of computational statistics to large collections of texts and images. Of particular interest in his research is the study of multimodal data, such as newspapers, film, and television. He is the co-author of Humanities Data in R: Exploring Networks, Geospatial Data, Images and Texts (Springer, 2015) and A Computational Approach to Statistical Learning (CRC Press, 2019).

Lauren Tilton is Assistant Professor of Digital Humanities in the Department of Rhetoric & Communication Studies and Research Fellow in the Digital Scholarship Lab (DSL) at the University of Richmond. Her research focuses on 20th and 21st century U.S. visual culture. She is director of Photogrammar, a digital public humanities project mapping New Deal and World War II documentary expression funded by the ACLS and NEH, and co-author of Humanities Data in R: Exploring Networks, Geospatial Data, Images and Texts (Springer, 2015). Her scholarship has appeared in journals such as American Quarterly, Archive Journal, Digital Humanities Quarterly, and Digital Scholarship in the Humanities. Her most recent project, Distant Viewing, focuses on large scale image analysis using computer vision and is funded by an NEH ODH Level II Advancement Grant. She also serves on the Association for Computing in the Humanities (ACH) Executive Council. She received her Ph.D. in American Studies from Yale University.