Microsoft Releases Florence-2: A Novel Vision Foundation Model With A Unified, Prompt-based Representation For A Variety Of Computer Vision And Vision-language Tasks

Microsoft researchers have developed a novel vision foundation model, Florence-2, which uses a unified, prompt-based representation to tackle various computer vision and vision-language tasks. This approach addresses challenges such as the need for a consistent architecture and limited data by creating a single representation for all vision activities.