Skip to main content
Localize objects in an image with LocateAnything-3B.

Parameters

Union[str, Image.Image, np.ndarray]
required
RGB image.
str
required
“detect” (open-vocabulary detection over comma-separated categories), “ground” (locate a referring phrase), or “point”.
str
default:"''"
Categories (detect) or referring phrase (ground/point).
float | None
Optional HTTP timeout.

Returns

Dict[str, Any]: Backend response.
  • detect/ground: {“output”: {“boxes”: [[x1,y1,x2,y2], …], “labels”: […], “scores”: […]}}
  • point: {“output”: {“points”: [[x, y], …]}}

Example