On Wednesday, Meta, the parent company of Facebook, announced that it had released a new AI model, along with the largest dataset of image annotations ever created. The model, known as the Segment Anything Model (SAM), can identify objects within images and videos, even if it has not encountered those items in its training. SAM allows objects to be selected by clicking on them or by writing text prompts. In a demonstration, writing the word “cat” prompted the tool to draw boxes around each of several cats in a photo.
The SAM model is already being used internally by Meta for activities such as photo tagging, moderating prohibited content, and determining which posts to recommend to Facebook and Instagram users. By releasing SAM and the accompanying dataset, Meta aims to broaden access to this type of technology.
Furthermore, Meta has been teasing a number of features that employ generative AI, such as a tool that creates surreal videos from text prompts and another that generates illustrations for children’s books from prose. CEO Mark Zuckerberg has said that incorporating such “creative aids” into Meta’s apps is a priority for this year.
The SAM model and dataset are available for download under a non-commercial license. Users who upload their own images to the accompanying prototype must agree to use it only for research purposes.
About Author
