feat: 🚀 initial keypoint support for transformers added

supervision/config.py

@@ -1,2 +1,3 @@
 CLASS_NAME_DATA_FIELD = "class_name"
 ORIENTED_BOX_COORDINATES = "xyxyxyxy"
+DESCRIPTORS_FIELD = "descriptors"

supervision/keypoint/core.py

@@ -7,7 +7,7 @@
 import numpy as np
 import numpy.typing as npt
 
-from supervision.config import CLASS_NAME_DATA_FIELD
+from supervision.config import CLASS_NAME_DATA_FIELD, DESCRIPTORS_FIELD
 from supervision.detection.utils import get_data_item, is_data_equal
 from supervision.validators import validate_keypoints_fields
 
@@ -509,6 +509,71 @@ def from_detectron2(cls, detectron2_results: Any) -> KeyPoints:
         else:
             return cls.empty()
 
+    @classmethod
+    def from_transformers(cls, transformers_results: List) -> KeyPoints:
+        """
+        Create a `sv.KeyPoints` object from the
+        [Transformers](https://huggingface.co/transformers/) inference result.
+
+        Args:
+            transformers_results (Any): The output of a
+                Hugging Face Transformers model containing instances with prediction data.
+
+        Returns:
+            A `sv.KeyPoints` object containing the keypoint coordinates, class IDs,
+                and class names, and confidences of each keypoint.
+
+        Example:
+            ```python
+            import cv2
+            import torch
+            from PIL import Image
+            import supervision as sv
+            from transformers import AutoImageProcessor, SuperPointForKeypointDetection
+
+            processor = AutoImageProcessor.from_pretrained("magic-leap-community/superpoint")
+            model = SuperPointForKeypointDetection.from_pretrained("magic-leap-community/superpoint")
+
+            image = cv2.imread(<SOURCE_IMAGE_PATH>)
+            image_pil = Image.fromarray(image)
+            inputs = processor(images,return_tensors="pt").to(model.device, model.dtype)
+            outputs = model(**inputs)
+            keypoints = sv.KeyPoints.from_transformers(outputs)
+            ```
+        """  # noqa: E501 // docs
+
+        keypoints_list: List[np.ndarray] = []
+        scores_list: List[np.ndarray] = []
+        descriptors_list: List[np.ndarray] = []
+        data: Dict[str, Any] = {}
+
+        for result in transformers_results:
+            if "keypoints" in result:
+                keypoints = result["keypoints"].detach().numpy()
+                scores = result["scores"].detach().numpy()
+
+                if keypoints.size > 0:
+                    keypoints_list.append(keypoints)
+                    scores_list.append(scores)
+
+            if "descriptors" in result:
+                descriptors = result["descriptors"].detach().numpy()
+
+                if descriptors.size > 0:
+                    descriptors_list.append(descriptors)
+
+        if not keypoints_list:
+            return cls.empty()
+
+        if descriptors_list:
+            data[DESCRIPTORS_FIELD] = np.array(descriptors_list)
+
+        return cls(
+            xy=np.array(keypoints_list, dtype=np.float32),
+            confidence=np.array(scores_list, dtype=np.float32),
+            data=data if data else {},
+        )
+
     def __getitem__(
         self, index: Union[int, slice, List[int], np.ndarray, str]
     ) -> Union[KeyPoints, List, np.ndarray, None]: