Meta has created an AI model called "Segment Anything," which can identify objects in images and videos without prior training, even detecting items not listed in the original data set.
This eliminates the need for pre-existing data sets and enables more precise object recognition, reducing the requirement for further training, which can increase efficiency.
However, the model may experience slow performance with complex image processing, and specialised AI tools are still needed for specific domains.
Meta will make both the AI model and dataset available for download under a non-commercial license, promoting research and enhancing accessibility to technology.