Meta presented and launched new computer vision tools that can be used to identify objects in an image.
Source: motionarray.com
These tools include a new Segment Anything (SAM) model and a set of Segment Anything masks with a volume of 1 billion units (SA-1B). The relevant information is contained in the publication, which was posted on the company’s blog last Wednesday, April 5.
The names of the tools were compiled on the basis of segmentation, which is the process of determining the pixels of an image belonging to a specific object.
Access to SAM is provided under a permissive open license. SA-1B is available for research purposes.
SAM has a general idea of what objects are. This means that this tool can identify those objects of the material world that previously did not fall into its field of view. SAM is characterized by generalized-abstract awareness of the very concept of an object. This level of understanding of the surrounding objective reality provides a wide range of use cases. The instrument does not require additional training to work in a previously unknown environment, for example, in cell microscopy or underwater space.
In a publication posted on the company’s blog, it is noted that this model can be used to help large artificial intelligence systems get a more voluminous and large-scale view of the world. The technology will also provide content creators with the ability to extract parts of an image when creating collages or editing videos. The use of tools for scientific researchers is that they can be used to study and track animals or objects of non-animal origin.
The SA-1B dataset includes 1.1 billion masks that cover various regions, levels of material and financial well-being, and demographic information.
SAM has the same level of effectiveness regardless of the social group in which it is used and the complexity of the situation.
Meta expects its tools to accelerate research into segmentation and a more general understanding of images and videos.
Computer vision, which is one of the fields of artificial intelligence, is used by companies in different ways. For example, retail technology company Pensa used computer vision and artificial intelligence to scan shelves.
Google uses computer vision, artificial intelligence, and billions of images to create highly accurate images of the terrain presented in its immersive representation on maps.
As we have reported earlier, Meta Expands ID Verification Subscription Service to US.
