Text SAM is an open-source model that uses free-form text prompts to extract various features by combining Grounding DINO and the Segment Anything Model (SAM). Grounding DINO acts as an open-set object detector, identifying objects based on text input. The detected objects are represented by bounding boxes, which are then passed to SAM. SAM segments the objects within these bounding boxes or specified points, generating precise masks. These masks are subsequently converted into polygons and returned as GIS features. The extracted features correspond to the objects described in the input text prompts and can include vehicles, swimming pools, ships, airplanes, solar panels, and more.