[ad_1]
Here is a place where you probably did not expect machine learning, but we are there anyway.
Google revealed "Move Mirror" in a blog on July 19th. It's an automatic learning experience that matches your poses from other people in the same pose.
The reason for its existence? Fun, mainly. Google also wanted to "make machine learning more accessible to coders and designers" while encouraging them to use the technology and use it for their own applications.
The "mirror" uses Google's open-source "pose estimation model" called PoseNet, which detects body poses, and TensorFlow.js, a library for the learning setting. machine in the browser.
In the search for a corresponding image, the experience uses your "pose information" – the location of 17 different parts of the body, including the shoulders, ankles and hips. According to the Google Explainer, it does not take into account individual characteristics, such as gender, size or body type.
I gave him a real challenge dancing like a ball of madness and he seemed to continue throwing a young woman in a white dress.
Using computers to detect poses is not new, of course – motion capture technology has been used for decades to capture real human movements for blockbusters. The video games have also used, just look at the Microsoft 3D imaging device, the Kinect. The triumph is that everything happens in a browser, with just a webcam.
Google does not send any of your images to its servers, all image recognition is done locally, in a browser. Technology also does not recognize who is in the picture because there is "no identifiable personal information badociated with the pose's estimate."
If you are interested in the Amazing amount of work needed to create Move Mirror, TensorFlow broad overview of the challenges and obstacles of programming that they have overcome in his blog.
You can try it for yourself, provided you have a webcam, on the Google Experiments page.
'Hello, Humans': Google's Duplex could make the AI badistant the liveliest to date
Comic-Con 2018: We're on the road to America's epic entertainment geekfest, and bringing you all the latest.
Source link