image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB) # Convert BGR -> RGB

image.flags.writeable = False # Set the image to read-only mode

results = model.process(image) # MediaPipe Model Prediction

image.flags.writeable = True # Set the image back to writeable mode

image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR) # Convert back to RGB -> BGR

mediapipe_drawing.draw_landmarks(image, results.face_landmarks, mediapipe_holistic.FACEMESH_TESSELATION,

mediapipe_drawing.DrawingSpec(color=(255,0,0), thickness=1, circle_radius=1), # Dot Color/Size

mediapipe_drawing.DrawingSpec(color=(0,255,0), thickness=1, circle_radius=1)) # Line Color/Size

mediapipe_drawing.draw_landmarks(image, results.pose_landmarks, mediapipe_holistic.POSE_CONNECTIONS,

mediapipe_drawing.DrawingSpec(color=(255, 0, 0), thickness=2, circle_radius=3), # Dot Color/Size

mediapipe_drawing.DrawingSpec(color=(0, 255, 0), thickness=2, circle_radius=2)) # Line Color/Size

mediapipe.solutions.drawing_utils.draw_landmarks(image, results.left_hand_landmarks, mediapipe_holistic.HAND_CONNECTIONS,

mediapipe_drawing.DrawingSpec(color=(255, 0, 0), thickness=2, circle_radius=3), # Dot Color/Size

mediapipe_drawing.DrawingSpec(color=(0, 255, 0), thickness=2, circle_radius=2)) # Line Color/Size

mediapipe.solutions.drawing_utils.draw_landmarks(image, results.right_hand_landmarks, mediapipe_holistic.HAND_CONNECTIONS,

mediapipe_drawing.DrawingSpec(color=(255, 0, 0), thickness=2, circle_radius=3), # Dot Color/Size

if results.face_landmarks:

face = np.array([[res.x, res.y, res.z] for res in results.face_landmarks.landmark]).flatten() # Flatten into 1-D array

else:

face = np.zeros(468 * 3) # Fill with zeros if no face detected

if results.pose_landmarks:

pose = np.array([[res.x, res.y, res.z, res.visibility] for res in results.pose_landmarks.landmark]).flatten() # Flatten into 1-D array

else:

pose = np.zeros(33 * 4) # Fill with zeros if no pose detected

if results.left_hand_landmarks:

left_hand = np.array([[res.x, res.y, res.z] for res in results.left_hand_landmarks.landmark]).flatten() # Flatten into 1-D array

else:

left_hand = np.zeros(21 * 3) # Fill with zeros if no left hand detected

if results.right_hand_landmarks:

right_hand = np.array([[res.x, res.y, res.z] for res in results.right_hand_landmarks.landmark]).flatten() # Flatten into 1-D array

else:

right_hand = np.zeros(21 * 3) # Fill with zeros if no right hand detected

keypoints = np.concatenate([face, pose, left_hand, right_hand]) # Concatenate into one array

#keypoints = np.concatenate([pose, left_hand, right_hand]) # Concatenate into one array

#keypoints = np.concatenate([left_hand, right_hand]) # Concatenate into one array

img_pil = Image.fromarray(src)

draw = ImageDraw.Draw(img_pil)

font = ImageFont.truetype('fonts/gulim.ttc', font_size)

draw.text(pos, text, font=font, fill= font_color)

# Smoothly transition between two sets of keypoints over a number of frames.

transition_frames = []

for t in range(1, num_transition_frames + 1):

alpha = t / (num_transition_frames + 1) # Interpolation factor

interpolated_keypoints = keypoints1 * (1 - alpha) + keypoints2 * alpha

transition_frames.append(interpolated_keypoints)

# Load all sequences and frames for a word.

all_frames = []

for sequence in sorted(os.listdir(word_path), key=lambda x: int(x)):

sequence_path = os.path.join(word_path, sequence)

if not os.path.isdir(sequence_path):

continue

for frame_file in sorted(os.listdir(sequence_path), key=lambda x: int(x.split('.')[0])):

frame_path = os.path.join(sequence_path, frame_file)

all_frames.append(np.load(frame_path))

words = []

with open(file_path, 'r', encoding='utf-8') as file:

for line in file:

# Split the line by '=' and extract the word (strip quotes and spaces)

if '=' in line:

_, word = line.split('=')

words.append(word.strip().strip('"'))