#'MapAnything

'MapAnything is a unified transformer model that directly outputs factored metric 3D scene geometry from images and optional sensor inputs, achieving state-of-the-art results across multiple reconstruction tasks and datasets.'