[12:18 Sun,9.April 2023 by Thomas Richter] |
Meta has released the extremely interesting "Segment Anything" (SAM) AI for creating masks and has also made it freely available on the web. "Segment Anything" recognizes objects in images and automatically exposes them as individual masks. Segmentation means recognizing which pixels in an image belong to an object - a task that seems quite simple to us as humans, but has always been a major challenge for computer programs. The new machine learning AI model has been trained on millions of images including object segmentations and has thus learned a general idea of what objects are. It can now generate masks for any object in any image or video - even for objects that were not present in the training material.
![]() Try it yourselfThere are different masking modes: in interactive mode you can click on image objects with the mouse and they will be masked automatically - also you can define an image area with a rectangle and all objects in it will be masked; in "all" mode every object in an image will be assigned a mask. Fortunately Meta offers a ![]() ![]() Image segmented SAM&s ability to recognize objects in images and videos could have many applications: in creative work for simply masking objects in photos and videos for targeted post-processing or tracking. In AI research, SAM could be used as part of more complex systems to gain a more general multimodal understanding of the world, e.g., to understand both the visual and textual content of a web page. Other uses include (traffic) surveillance systems, object recognition in spaces e.g. for robots, and many other applications. ![]() Dataset "Segment Anything" consists of the general Segment Anything Model (SAM) as well as the Segment Anything ![]() ![]() ![]() deutsche Version dieser Seite: Neue freie Meta KI erstellt automatisch Masken für alle Objekte in Fotos oder Videos |
![]() |