FAQs
Is VizPol available to the general public?
No. At this point, VizPol is available only to select journalists and researchers who are working on related projects (e.g. covering in-person events, or researching political representation and imagery). If this describes your work and you would like a mobile account for VizPol, please fill out this form. If you are a journalist, you will be asked to submit a link to an online portfolio/social media profile, as well as links to two works of journalism that have been published by independent news organizations. If you are a researcher, you will be asked to provide a link to an institutional profile and two published works of related research. If you have questions or would like to discuss institutional/web app access, please reach out to us via email at vizpolresearch@gmail.com.
What data does the VizPol app collect?
Our research team takes the privacy, security and intellectual property rights of our users extremely seriously, and our research team has a deep understanding of the ways that data collected by apps can be misused. As a result, the VizPol mobile apps do not collect any data apart from what users actively submit. In other words, we do not collect device IDs, location information or other potentially identifying metadata— only the images that users submit when they look up a symbol in a photo. Even then, user-submitted images are only used to improve our database if that option is explicitly selected. Finally, any images submitted to our database are put through a face-blurring algorithm before they are ever saved to disk, in order to prevent even incidental identification of individuals in the database.
How do new symbols/graphics get added to the database?
VizPol is a “human-in-the-loop” system: while a combination of computer vision and machine-learning “AI” does the work of matching the symbols in submitted images to the ones in our database, the job of adding new symbols is done by a human. This gives us a chance to research new symbols both to make sure they make sense to add to the database (our goal is not for the app to recognize all political graphics, just less-recognizable ones), and to make sure the new entry has some appropriate context attached to it.
How does the app actually work?
The functioning of the app is very simple: with the app installed, select a photo from your phone’s camera roll, or take a photo from within the app. The app will prompt you to zoom in on the symbol of interest, and then you simply click “Identify symbol”. If a very high confidence match is found, the app will return only that option. Otherwise, it will return the three closest matches in the database. After that you simply click on the correct symbol, or choose “Not my symbol”. You can then choose (and we hope you will!) to submit your image to the database to improve the tool. That’s it!
[insert flow chart here]
