Meta’s Segment Anything AI can detect objects without training

  • Meta has published a version of its Segment Anything AI model that can detect objects that are not a part of its training set.
  • The AI can detect objects in both photos and videos.
  • Meta adds that it is able to detect objects based on prompts that it is giving, highlighting said object.

It seems like everyone and their dog is obsessed with AI at the moment, with OpenAI’s ChatGPT platform leading the way. We have also seen the likes of Microsoft and Google enter the fray, but now Meta has detailed its own AI efforts with a new model of Segment Anything tool made available to developers.

What makes this model particularly impressive is the fact that it is able to detect objects that it has not been trained on, according to Meta.

This extends to both photos and videos, with Reuters also pointing out the this version of Segment Anything (SAM) is able to highlight objects when prompted via text. To that end, it you input dog as a search term, the model will highlight all dogs, if any, in a piece of media.

Added to this is the ability to collaborate with, and be integrated into other models.

“SAM has learned a general notion of what objects are, and it can generate masks for any object in any image or any video, even including objects and image types that it had not encountered during training. SAM is general enough to cover a broad set of use cases and can be used out of the box on new image ‘domains’ — whether underwater photos or cell microscopy — without requiring additional training (a capability often referred to as zero-shot transfer),” explains Meta in a blog post.

“In the future, SAM could be used to help power applications in numerous domains that require finding and segmenting any object in any image. For the AI research community and others, SAM could become a component in larger AI systems for more general multimodal understanding of the world, for example, understanding both the visual and text content of a webpage,” it adds.

At the time of writing, it remains to be seen where Segment Anything will be applied in the real world, but it certainly looks like Meta is making serious strides when it comes to object detection in AI.

“We believe the possibilities are broad, and we are excited by the many potential use cases we haven’t even imagined yet,” it concluded.


About Author


Related News