Last Wednesday (5), Meta announced the Segment Anything Model (SAM), its new artificial intelligence (AI) that can identify individual objects in images and videos.
The SAM identifies which image pixels belong to a given object. In this way, in a demonstration video, he showed that, when writing the word “cat”, the tool drew boxes around each of the several cats present in an image.
see more
How to get your CNH for free in 2023?
After hacker attacks, Microsoft releases free tools for…
In another example, while filming, the AI identified each object that appeared on the recording screen.
Another novelty is that the tool can also separate the same object by layers or highlight them in 3D format.
Meta’s AI team, in an official statement, explained more details:
”We intend to democratize segmentation by introducing the Segment Anything project: a new task, dataset and model for image segmentation, as explained in our work on search. We are releasing our general Segment Anything (SAM) model and our Segment Anything 1-Billion (SA-1B) mask dataset, the largest dataset of all time, to enable a broad set of applications and to promote further research on underlying models for vision computational.”
Still according to Meta, the features are based on the type of generative AI popularized by ChatGPT, which creates entirely new content rather than simply identifying or categorizing data. Mark Zuckerberg, Meta's chief executive, said incorporating these "creative aids" into the company's applications is a priority for this year.
Meta already uses SAM-like technology internally for activities like photo tagging, moderation of prohibited content and determining which posts to recommend to Facebook and Instagram users. However, the release of SAM to the public and researchers aims to expand access to this type of technology.
The SAM model and dataset are now available for download under a non-commercial license. Test the SAM in the link https://segment-anything.com/.
Lover of movies and series and everything that involves cinema. An active curious on the networks, always connected to information about the web.