The AI-based Segment Anything Model (SAM) from Meta can successfully recognize and separate particular things in photos and movies. The program is free and open source, and it provides a more sophisticated selecting tool than competitors like Google’s Magic Eraser. SAM is referred to as a “foundation model” for both interactive and automatic image segmentation and is defined as being able to output various masks even when there is uncertainty regarding the object. The largest segmentation dataset ever is the SA-1B dataset from Meta, which contains information on 1 billion masks. With an average of 100 masks per image, the system was trained on 11 million photos.
SAM’s open sourcing is a huge step, but Meta has been concentrating more on AI and its possibilities for commercial use, such as generative AI for advertisements. In addition, the business is preparing a public release for LLaMA, a ChatGPT rival. The potential for misuse of the technology, such as facial recognition, highlights the need for regulation in the industry, even though SAM may be employed in AR or VR applications.