Module `src.recognition.hand`

Classes

class Hand (gesture_backlog: int = 20, threshold: float = 0.8)

Hand - a wrapper for processing hand landmarks with persistence.

Since lots of different features need access to the GestureClassifier, hand ensures that a landmark list is only processed once.

After a gesture for the hand landmarks has been estimated, it is passed on to the persistence checker.

gesture_backlog: How many gestures (queue size) form the basis for persistence checks.
threshold: Minimum ratio ([0.0, 1.0]) for regarding the current gesture as intentional.

var has_intent: Property that signifies if the object currently holds a gesture intended as intentful. From outside the class, this property is readable only

def update_state(self, landmarks: mediapipe.framework.formats.landmark_pb2.NormalizedLandmarkList) ‑> NoneType

Updates the state of the hand.

Takes the provided landmarks to estimate a gesture and passes this on to the persistence checker.

landmarks: A list of landmarks for one hand. Also accepts None, which is equivalent to no gesture being found.