Visual computational sociology : computer vision methods and challenges
Abstract/Contents
- Abstract
- Targeted social and economic policies require an understanding of a country's demo- graphic makeup. Thus, countries like the United States spend more than 1 billion dollars a year gathering survey based census data such as race, education, occupation and unemployment rates. In this work, we explore the use of publicly available digital imagery to measure demographics without the high costs incurred by surveys. These results can then be inputs to critical socioeconomic policies. In particular, we apply computer vision methods to 50 million Google Street View images from 200 cities and show that a wide range of characteristics such as income, education levels, voting patterns, race, CO2 emission per capita and other market research results can be accurately inferred from these images. We specifically use fine-grained image recognition techniques to detect and classify approximately 21.8 million cars and use them to predict demographics. One can imagine using multiple objects such as trees, clothes, houses or people themselves to further extend this analysis and improve its accuracy. However, scal- ing these methods to other objects such as trees, clothes, houses or other objects using current supervised methods is infeasible. Fine-grained datasets are notoriously expensive to collect and it is not possible to annotate images of objects in every pos- sible form they can appear in the world. We present scalable methods of collecting synthetic fine-grained datasets and present an attribute based fine-grained domain adaptation method to perform data efficient classification.
Description
| Type of resource | text |
|---|---|
| Form | electronic; electronic resource; remote |
| Extent | 1 online resource. |
| Publication date | 2017 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Associated with | Gebru, Timnit |
|---|---|
| Associated with | Stanford University, Department of Electrical Engineering. |
| Primary advisor | Li, Fei Fei, 1976- |
| Thesis advisor | Li, Fei Fei, 1976- |
| Thesis advisor | Ermon, Stefano |
| Thesis advisor | Gill, John T III |
| Advisor | Ermon, Stefano |
| Advisor | Gill, John T III |
Subjects
| Genre | Theses |
|---|
Bibliographic information
| Statement of responsibility | Timnit Gebru. |
|---|---|
| Note | Submitted to the Department of Electrical Engineering. |
| Thesis | Thesis (Ph.D.)--Stanford University, 2017. |
| Location | electronic resource |
Access conditions
- Copyright
- © 2017 by Timnit W Gebru
- License
- This work is licensed under a Creative Commons Attribution Non Commercial No Derivatives 3.0 Unported license (CC BY-NC-ND).
Also listed in
Loading usage metrics...