Meta announced the release of an artificial intelligence model that can identify specific objects in images. In addition to the model, Meta also publishes a dataset of image annotations, which it claims is the most extensive to date.
Meta announced the release of an artificial intelligence model that can identify specific objects in images. In addition to the model, Meta also publishes a dataset of image annotations, which it claims is the most extensive to date.
According to a recent blog post by Meta’s research division, the company has developed an advanced object recognition model called the Segment Anything Model (SAM). SAM is designed to identify objects in images and videos even if they were not encountered during the training phase. In this model, users can select an object by clicking on it or using a text prompt such as the word “cat”. In a demonstration, SAM was able to accurately draw boxes around multiple cats in a picture in response to written prompts.
According to a recent blog post by Meta’s research division, the company has developed an advanced object recognition model called the Segment Anything Model (SAM). SAM is designed to identify objects in images and videos even if they were not encountered during the training phase. In this model, users can select an object by clicking on it or using a text prompt such as the word “cat”. In a demonstration, SAM was able to accurately draw boxes around multiple cats in a picture in response to written prompts.
Meta uses SAM-like technology internally for tasks such as tagging photos, moderating banned content, and recommending posts to Facebook and Instagram users. The company says the release of SAM will expand access to this type of advanced technology beyond its internal operations.
Meta uses SAM-like technology internally for tasks such as tagging photos, moderating banned content, and recommending posts to Facebook and Instagram users. The company says the release of SAM will expand access to this type of advanced technology beyond its internal operations.
The company has made the SAM models and datasets available for download under a non-commercial license. However, those who upload their own images to the accompanying prototype must agree to use the tool for research purposes only.
The company has made the SAM models and datasets available for download under a non-commercial license. However, those who upload their own images to the accompanying prototype must agree to use the tool for research purposes only.
“In the future, SAM can be used to power applications in many domains that need to find and segment any object in any image. It could be a component of larger AI systems for general multimodal world understanding (e.g., understanding both visual and textual content of web pages).In the AR/VR domain, SAM can be It allows you to select an object based on the user’s gaze and ‘lift’ it into 3D,” Meta said in a blog post.
“In the future, SAM can be used to power applications in many domains that need to find and segment any object in any image. It could be a component of larger AI systems for general multimodal world understanding (e.g., understanding both visual and textual content of web pages).In the AR/VR domain, SAM can be It allows you to select an object based on the user’s gaze and ‘lift’ it into 3D,” Meta said in a blog post.
The tech giant suggests that SAM could have some applications for content creators, such as the ability to extract image regions for use in collages and video editing. Additionally, the model could be useful in scientific research, allowing scientists to locate and track animals and objects of interest within video footage of natural phenomena on Earth or in space.
