Launch: Use Florence-2 in Roboflow Workflows

Florence-2, a cutting-edge multimodal model developed by Microsoft and licensed under MIT, is integrated into Roboflow Workflows, a web-based computer vision application builder. The model supports a variety of tasks, including zero-shot object detection and optical character recognition (OCR), allowing users to build applications quickly without writing code. Users can create a Workflow in Roboflow to utilize Florence-2 for tasks such as object detection and segmentation by setting up a dedicated deployment with a GPU or using a local inference server. The platform facilitates these operations through a series of configurable blocks within the Workflow, enabling the visualization of results using bounding boxes or masks. Notably, Florence-2 can run offline on users' hardware, eliminating the need for external API calls, which enhances privacy and control. The guide provides detailed instructions on setting up and testing Workflows using Florence-2 for both object detection and OCR, with examples illustrating the practical application of these capabilities.