Add to Favourites
To login click here

Scientists at MIT and Adobe Research have developed a technique that can accurately identify all pixels in an image representing a given material, even when objects have varying shapes and sizes and the machine-learning model is tricked by shadows or lighting conditions. The method can be used for videos and has applications in scene understanding for robotics, image editing, web recommendation systems, and more. The technique could facilitate the selection of all the other pixels in an image that are made from the same material.