These days, there's a widespread belief that governments are spying on people using the cameras built into their laptops. That's why many people, including Mark Zuckerberg, tend to cover their cameras when they're not using them.
However, a Google web engineer , Jason Mayes, has discovered another way to fool those who spy on us.
Mayes has developed an AI-powered tool that can remove people from a live webcam stream without affecting the rest of the footage.
To create the tool, Mayes used TensorFlow and Javascript to be able to extract people from complex backgrounds in real time. TensorFlow is a free and open-source framework offered by Google for building machine learning models.
The “disappearing people” tool uses machine learning to learn over time how to create artificial static background images that do not include the people in a space. The neural network is trained to recognize human beings and then “disappear” them from a live webcam.
In the demonstration presented by the engineer, we can see that the tool is not 100% perfect yet, as some parts of its body are still visible in the video of it entering the frame. However, we hope that the tool will improve as time goes by and it is trained better and better.
Mayes has published the source code of the tool on Github and you can try it out by using the code in your browser. What's most exciting about the toolis the fact that it works in real time and can be used to, for example, make people disappear from a surveillance camera.
