Meta launches AI model that can identify items in images – 4/5/2023 – Tech

Meta released this Wednesday (5) an artificial intelligence model that can identify individual objects within an image. According to the company, the tool is the largest of its kind ever created.

The company’s research division said its Universal Segmentation Model (SAM) can identify objects in images and videos even in cases where it didn’t find those items in its training.

Using SAM, objects can be selected by clicking on them or by writing text commands. In a demo, when the user wrote the word “cat”, the tool drew boxes around each of the cats in a photo.

Big tech companies have been touting their advances in artificial intelligence ever since OpenAI’s ChatGPT became popular, triggering a wave of investment and a race to dominate the space.

Meta has introduced several features that deploy the type of generative AI popularized by ChatGPT, which creates content rather than simply identifying or categorizing data like other AI systems, although it has yet to release a product.

Examples include a tool that generates surreal videos from text commands and another that generates children’s book illustrations from the story.

Meta Chief Executive Mark Zuckerberg, who previously centered the company around what he called the “metaverse,” said incorporating these “creative aids” of generative AI into the company’s applications is a priority this year.

Meta already uses SAM-like technology internally for things like photo tagging, moderating banned content, and determining which posts it can recommend to Facebook and Instagram users.