A machine learning software system observed government officials and showed which of them were actually paying attention. To top it off, the software showed the politicians’ distraction percentage.
Dries Depoorter used machine learning to observe government officials
It is often easy to tell in person if a person is distracted or not. However, when it comes to videos of multiple people, it can be quite difficult to pay attention to each person in the video. However, this is something a newly built software system was actually able to do.
According to the story of Hackaday, Dries Depoorter is known for his talent for highly technical projects with a strong artistic touch. The latest piece is no exception as it uses machine learning and takes it to the next level.
Live streaming sessions viewed by the software system via Python and machine learning
Dries Depoorter created the Flemish Scrollers, which is a software system that watches the Flemish government’s live streaming sessions. Through its use of Python and machine learning, the software is able to identify and highlight politicians who pull out their phones and march.
The results were shared live on Instagram and Twitter for the world to see.
The project actually started in July 2021 and has been working dutifully ever since.
Other important contributions of technology
The project also has another contribution besides showing the distraction percentage of the subjects in the video. According to the Hackaday article, the project can also be seen as a good example of how confidence in results can be properly managed depending on the particular application.
In this particular case, false negatives, such as a politician using a phone but the software was unable to pick it up correctly, are actually much more acceptable than false positives. False positives include instances where a member is incorrectly identified or even wrongly called out for using a mobile device even though they are not using any device.
Read also : Can students still learn from quick-view video lectures? A UCLA study says it’s possible
Facial recognition and object detection with Keras
For those interested not only in the project but also in the concept behind the project, there are open source facial recognition resources already available online. one of which is Kerasknown as an open source software library used for object detection as well as facial recognition.
Interested persons can find out more about Keras via its GitHub deposit. Additionally, Hackaday’s article notes that they’ve seen it used in a number of different things like bat detection as well as automatic waste provisioning.
Hackaday’s article notes that, for those interested in machine learning applications, Keras might be a good place to start.
Every meeting of the Flemish government in Belgium is broadcast live on a YouTube channel.
The software searches for phones and tries to identify a distracted politician. This with the help of AI. The results are then posted on social media with the politician tagged. By @driesdepoorter pic.twitter.com/QUQHiYx2HI— The Flemish Scrollers (@FlemishScroller) July 5, 2021
Related article: The most detailed 3D map of the universe to arrive, thanks to DESI-7.5 million galaxies already mapped
This article belongs to Tech Times
Written by Urian B.
ⓒ 2021 TECHTIMES.com All rights reserved. Do not reproduce without permission.
