Meta releases AI model that can detect objects in images

Meta, the company that owns Facebook, has announced the release of a new artificial intelligence model that can identify objects within an image.

– Key things to note

The model, Segment Anything Model (SAM), can detect individual objects in images and videos, even if it has not been trained on them before. Meta also released a dataset of image annotations, which it said was the largest of its kind.

SAM can be used to select objects in images by clicking on them or using text prompts. In one demonstration, writing “cat” prompted the tool to draw boxes around all the cats in a photo.

– Why it matters

Image segmentation has a multitude of applications, including photo editing, scientific image analysis, and larger AI systems that require a broad multimodal comprehension of the world. Notably, it can also be used for augmented reality (AR) and virtual reality (VR).

Previous segmentation models necessitated either interactive segmentation, requiring individual guidance, or automatic segmentation, requiring extensive manual annotation of objects for training.

With SAM, a single model, either segmentation method can be easily performed, eliminating the need for practitioners to gather their own segmentation data and fine-tune a model for their specific use case. This saves time and effort, making the segmentation process more efficient.

“SAM could also be used to aid scientific study of natural occurrences on Earth or even in space, for example, by localizing animals or objects to study and track in video. We believe the possibilities are broad, and we are excited by the many potential use cases we haven’t even imagined yet,” Meta said.

– Learn more

Big tech companies have been highlighting their breakthroughs in artificial intelligence since the Microsoft-backed OpenAI’s ChatGPT chatbot became popular.

Meta has teased the use of generative AI in its products, which creates new content instead of just categorizing data. CEO Mark Zuckerberg has stated that incorporating this type of AI into Meta’s apps is a priority for the company.

SAM is already used internally by Meta for tagging photos, moderating content, and recommending posts to users of Facebook and Instagram. The release of SAM will allow more people to access this technology.

The SAM model and dataset can be downloaded under a non-commercial license, and users must agree to use the accompanying prototype for research purposes only.