Meta Launches New AI Model That Can Recognize Details in Photos
Tùng Anh•07/04/2023
On April 5, Meta - Facebook's parent company - introduced an artificial intelligence (AI) model capable of recognizing every detail in an image, along with an image captioning dataset that the company said is the largest ever in this field.
Meta icon on a phone screen in Moscow, Russia. Photo: AFP/TTXVN
In a blog post, Meta’s research division said that its Segmentation Model (SAM) can recognize objects in photos and videos even if it has never been taught about them. Using SAM, users can select objects in a photo by clicking on them or writing a suggestion. For example, just typing the phrase “cat” will immediately draw a frame around each cat in the photo. Meta’s new AI model comes as many tech giants are racing to dominate this field. The Facebook-owned company is also internally testing several features based on Generative AI — a type of AI that focuses on creating entirely new content, rather than just recognizing or classifying data like other AIs. One popular Generative AI tool today is ChatGPT. Meta CEO Mark Zuckerberg said integrating AI creative support tools into Meta's apps is a top priority for the company in 2023. The company currently uses SAM-like technology internally for a number of activities such as tagging photos, moderating content, and determining user recommendations on the Facebook and Instagram platforms.
Meta says the launch of SAM will expand access to the technology. Users will be able to download the SAM model and dataset for non-commercial use.
Comment (0)