The AI normally needs to be trained on the existing material to detect objects, but Meta has a way for the technology to detect items without help. The social media giant has published a “Segment Anything” AI model that can detect objects in images and videos even if they were not part of the training set. You can select items by clicking on them or by using free-form text prompts. As Reuters Explainyou can type the word “cat” and watch the AI highlight all the felines in a given photo.
The model can also work in conjunction with other models. It can help rebuild an object in 3D using a single image or draw from views from a mixed reality headset. Effectively, Segment Anything can limit the need for additional AI training.
Both the AI model and the dataset will be available for download under a non-commercial license. That is, the creators cannot use it for products. This is primarily for research and expanding access to technology. Right now, Meta uses somewhat similar technology to moderate banned content, recommend posts, and tag photos.
The developers acknowledge that the existing model is flawed. It can miss finer details and is not as accurate at detecting boundaries as some models. And while Segment Anything can handle prompts in real time, it bogs down when it comes to demanding image processing. Some more specialized AI tools are likely to outperform this model in their respective fields, says Meta.
You won’t see this AI in robots or other devices where fast and accurate object detection is (usually) vital. However, models like this can still help in situations where relying solely on training data is impractical. A social network could use technology to keep up with a rapidly growing volume of content. If nothing else, this shows that Meta wants to generalize computer vision.
Meta is no stranger to sharing AI advancements, like a translator of unwritten languages. That said, there’s pressure on the company to prove it’s as much a powerhouse in the category as tech heavyweights like Google and Microsoft. It’s already planning generative AI “characters” for its social apps, and inventions like Segment Anything show it has some advantages of its own.
