SAN FRANCISCO: Facebook’s owner Meta published an (AI) artificial intelligence model on Wednesday capable of identifying individual objects from within an image, together with image annotations’ dataset that it stated was the biggest ever of its kind.
The company’s research section said in a blog post stated that its SAM (Segment Anything Model) could identify objects in videos and images even in instances where it had not been exposed to those items in its training.
With the help of SAM, items can be selected through clicking on them or writing prompts in text. In one presentation, after the word “cat” was written, the tool began to draw boxes by encircling each of a number of cats in a photo.
AI breakthroughs being trumpeted by tech giant
Major tech companies have been trumpeting their AI breakthroughs since OpenAI’s ChatGPT chatbot, which has Microsoft’s backing, became a sensation last fall, leading to a flurry of investments and a race to dominate the sector.
Meta has shown numerous features that utilize generative AI of the type popularized by ChatGPT, which is able to create completely new content instead of categorizing data or simply identifying like other AI, although it is yet to release a product.
Examples include a function that uses text prompts and conjures up surrealist videos and another tool that can generate kids’ book illustrations from writing in prose.
Meta’s Chief Executive Mark Zuckerberg has stated that adding these types of generative AI “creative aids” into the company’s apps is a priority this year.
