Computer Vision News - March 2022

8 Computer Vision Tool # Run MediaPipe Hands. with mp_hands.Hands( static_image_mode=True, max_num_hands= 2 , min_detection_confidence= 0.5 ) as hands: for name, image in images.items(): # Convert the BGR image to RGB, flip the image around y-axis for correct # handedness output and process it with MediaPipe Hands. results = hands.process(cv2.flip(cv2.cvtColor(image, cv2.COLOR_BGR2RGB), 1 )) # Print handedness (left v.s. right hand). print (f'Handedness of {name}:') print (results.multi_handedness) if not results.multi_hand_landmarks: continue # Draw hand landmarks of each hand. print (f'Hand landmarks of {name}:') image_hight, image_width, _ = image.shape annotated_image = cv2.flip(image.copy(), 1 ) for hand_landmarks in results.multi_hand_landmarks: # Print index finger tip coordinates. print ( f'Index finger tip coordinate: (', f'{hand_landmarks.landmark[mp_hands.HandLandmark.INDEX_FINGER_TIP].x * im f'{hand_landmarks.landmark[mp_hands.HandLandmark.INDEX_FINGER_TIP].y * im ) mp_drawing.draw_landmarks( annotated_image, hand_landmarks, mp_hands.HAND_CONNECTIONS, mp_drawing_styles.get_default_hand_landmarks_style(), mp_drawing_styles.get_default_hand_connections_style()) cv2.imshow(“Landmarks”, cv2.flip(annotated_image,1))