Facebook owner Meta released an artificial intelligence model that can detect individual items within an image, as well as the largest dataset of image annotations ever.
In a blog post, the company’s research division stated that its Segment Anything Model, or SAM, could recognise objects in photos and videos even when it had not met those items during training.
Meta new tech
Objects can be selected with SAM by clicking on them or typing text prompts. Writing the word “cat” prompted the tool to draw boxes around each of multiple cats in a photo in one instance.
Since Microsoft-backed OpenAI’s ChatGPT chatbot became a sensation in the autumn, igniting a flood of investments and a race to dominate the area, big tech companies have been trumpeting their artificial intelligence advances.
Although Meta has not yet developed a product, it has teased numerous features that use the type of generative AI popularised by ChatGPT, which creates brand-new content rather than merely detecting or categorising data like conventional AI.
A technology that makes surrealist movies from word prompts is one example, as is another that generates children’s book pictures from prose.
According to CEO Mark Zuckerberg, embedding such generative AI “creative aids” into Meta’s programmes is a focus this year.
Internally, Meta already use SAM-like technologies for tasks such as tagging photographs, censoring forbidden content, and choosing which posts to promote to Facebook and Instagram users. According to the corporation, the distribution of SAM will increase access to that type of technology.
To read our blog on “Facebook to fire thousands of employees,” click here
