Meta, formerly known as Facebook, has released a new model and dataset called “Segment Anything” to aid in the development of augmented reality (AR) experiences. The model aims to help identify different objects in a photo or video, providing functional insights overlaid on the screen. This technology will have applications in AI, VR, and content creation. Meta has released the Segment Anything Model (SAM) and the Segment Anything 1-Billion mask dataset (SA-1B), which is the largest-ever segmentation dataset. This release will enable developers to create a broad set of applications and foster further research into foundation models for computer vision. While the first iteration of Meta’s AR wearables was reportedly delayed due to cost-cutting measures, the release of the Segment Anything dataset points to its ongoing development in AR technology, which could see the release of another version of its smart glasses in the future.
