A meta-AI model that allows for the identification of things in images

(Reuters) The Facebook-owned Meta (META.O) released the data on Wednesday, revealing an artificial intelligence model that is able to determine the individual objects contained in an image. The model is also the largest collection of annotations for images ever released.

Segment Anything Model (SAM)

The company’s research division stated in a blog post its Segment Anything Model, or SAM could detect objects in videos and images even though it had not previously seen the objects.

With SAM objects, they can be selected by simply clicking on them or writing text-based prompts. In one example when you wrote”cat,” the phrase “cat” caused the tool to draw boxes on top of several cats.

After Microsoft’s backing (MSFT.O) the OpenAI ChatGPT chatbot went viral in the fall, triggering an influx of investment and a race to be the best in the market, major tech companies have been boasting about their advancements in artificial intelligence.

Segment Anything Model (SAM) Meta

While it’s yet to make the product accessible, Meta has teased a variety of features that utilize the artificially generated AI that was made famous via ChatGPT which creates completely new content instead of categorizing or detecting data as the traditional AI.

Examples include a program that makes surrealist video clips from text prompts and another program that converts prose into illustrations for books for children.

According to Chief Director Mark Zuckerberg, adding these”generative” AI “creative tools” to Meta’s products is the top goal for the coming year.

Internally, Meta makes use of SAM-like technology to accomplish tasks such as tagging photos as well as removing information that may be objectionable, and selecting which content to share with Facebook or Instagram users.

The company said it was confident that the launch of SAM will improve access to the technology.

A non-commercial license will be required to access the SAM model as well as the dataset. The users who post their own photos to the associated prototype must also agree to allow the use of their images solely for study purposes.

Leave a Comment

5 × 1 =

%d bloggers like this: