Ajit Karn
Media reported that Facebook parent company Meta has released an artificial intelligence (AI) model SAM that can identify the various objects within photographs.
The research division of Meta announced that it has published a Segment Anything Model (SAM) and the corresponding dataset to promote research into computer vision foundation models.
In its statement, Meta claimed that SAM is capable of recognizing objects within images and videos, even if it had not encountered them during its training.
“The model is designed and trained to be prompt so that it can transfer zero-shot to new image distributions and tasks,” explained Meta in a blog post.
“We evaluate its capabilities on a variety of tasks and find that its zero-shot performance is often comparable to or even better than previous fully supervised results,” the report continued.
Using SAM, it is possible to select objects by tapping on them or entering text prompts. media reports that in one demonstration, writing the word “cat” caused the tool to draw boxes around each of several animals in a photo.
Meta already uses comparable technology internally for activities such as tagging photos, moderating prohibited content, and determining which Facebook and Instagram posts to recommend to users. According to the company, the release of SAM would increase access to this sort of technology.
The SAM dataset and model will be downloadable under a non-commercial license. Users who contribute images to an accompanying prototype must also consent to its use solely for research purposes.
Also read this:RailTel receives a Rs 38.95 billion contract to provide lease line connectivity at Immigration Centres across India
Media reported that the release of the popular ChatGPT A.I. chatbot has catalyzed a race among tech titans to unveil similar tools, propelling AI to the forefront of national and international conversation in recent months.
