Meta recently decided to jump into the artificial intelligence pool. For quite a few months, when other technology companies were betting heavily on this field, it almost seemed that he was so focused on other fronts that he completely neglected this one. What’s more, until the end of February, the most transcendental thing in this regard was that Yann LeCun, the company’s head of AI, said that ChatGPT was not that innovative, a statement that should be rather subtle.

However, at the end of February released the flame (Large Language Model Meta AI), a model (not a service like ChatGPT, the new Bing and Bard, among others) that the company makes available to certain sectors through a non-commercial license. It was a big step forward for the company, but still, because its scope is quite limited, it hasn’t done much to promote its AI capabilities.

This is more important than it might seem at first glance. Let’s remember that the company had to face some pretty negative consequences as a result of its commitment to the Metaverse, a proposal as fantastic as it is fanciful, and that just over a year after it was presented in style, it seems doomed to be ostracized. The impact it had on the company, its accounts and its image was more than considerable, and considering that it was already low for a number of reasons, it is clear that you need more than ever to improve your image in any way you can.

This explains why Meta surprised us with a presentation today SAM, an artificial intelligence model capable of identifying different elements of an image. SAM (Search Anything Model), which is available to download and run on our own system, as well as a trial version through a website designed specifically for this purpose, is able to analyze images to identify each one. the elements (people, objects, etc.) that make them up.

For this purpose, the model was trained with a dataset of 1,000 million images, which Meta also shared, along with a fair amount of technical documentation for the dataset and model. On the test page, it shows its selection with 50,000 images on which we can check how SAM works, but also allows us to upload our own images to check how it works on them.

Feature detection in images is nothing new. For example, we’ve already talked about the magic eraser feature in Google Photos on some occasion. However, the meta has fully released SAM your reach will be much more global. In addition, features such as the one that allows you to “cut out” a selected element and use it separately later, exponentially expand their usefulness.