A machine learning software system watched government officials and showed which of them were actually paying attention. To top it off, the software showed the percentage of how distracted the politicians were.
Dries Depoorter Used Machine Learning to Observe Government Officials
It is often easy to tell in person whether a person is distracted or not. When it comes to videos of multiple people, however, it can be rather difficult to pay attention to each person on video. However, this is something that a newly built software system was actually able to do.
According to the story by Hackaday, Dries Depoorter is known for having a knack at highly technical projects with a solid artistic touch to them. The newest piece is no exception as it uses machine learning and takes it to the next level.
Software System Watched Live Streaming Sessions through Python and Machine LEarning
Dries Depoorter created the Flemish Scrollers, which is a software system that watches live streaming sessions of the Flemish government. Through its use of Python and machine learning, the software is capable of identifying and highlighting politicians who are pulling out their phones and scrolling.
The results were shared live on Instagram and Twitter for the whole world to see.
The project actually initially started back in July 2021 and has been dutifully running ever since then.
Other Important Contributions of the Technology
The project also has another contribution aside from showing the percentage of how distracted the subjects in the video are. According to the article by Hackaday, the project can also be considered a good example as to how confidence in results can be properly handled depending on the particular application.
In this particular case, false negatives, such as instances such as a politician using a phone but the software was not able to pick it up properly, are actually much more acceptable compared to false positives. False positives include instances where a member is incorrectly identified or even wrongly called-out for using a mobile device even if he or she is not using any device.
Read Also: Can Students Still Learn From Speed-Watching Video Lectures? UCLA Study Says It's Possible
Facial Recognition and Object Detection with Keras
For those interested in not just the project but also the concept behind the project, there are open-sourced facial recognition resources already available online. One of which is Keras, which is known as an open-source software library that is being used for object detection as well as facial recognition.
Interested individuals can learn more about Keras via its GitHub repository. In addition the article by Hackaday notes that they have seen it used in a number of different things like bat detection as well as automatic trash sourcing.
The article by Hackaday notes that, for those interested in machine learning applications, Keras could be a good place to start.
Every meeting of the flemish government in Belgium is live streamed on a YouTube channel.
— The Flemish Scrollers (@FlemishScroller) July 5, 2021
Software is searching for phones and tries to identify a distracted politician. This with the help of AI. The results are then posted to social with the politician tagged. By @driesdepoorter pic.twitter.com/QUQHiYx2HI
Related Article: Most Detailed 3D Map of the Universe To Arrive, Thanks to DESI-7.5 Million Galaxies Already Charted
This article is owned by Tech Times
Written by Urian B.
