uplauded utils & webcam.py

utils.py

@@ -0,0 +1,96 @@
+import math
+import numpy as np
+import mediapipe as mp
+from mediapipe.tasks import python
+from mediapipe.tasks.python import vision
+import cv2
+
+
+# landmarks of features from mediapipe
+face_points={
+"BLUSH_LEFT": [50],
+"BLUSH_RIGHT": [280],
+"LEFT_EYE": [33, 246, 161, 160, 159, 158, 157, 173, 133, 155, 154, 153, 145, 144, 163, 7, 33],
+"RIGHT_EYE": [362, 298, 384, 385, 386, 387, 388, 466, 263, 249, 390, 373, 374, 380, 381, 382, 362],
+"EYELINER_LEFT": [243, 112, 26, 22, 23, 24, 110, 25, 226, 130, 33, 7, 163, 144, 145, 153, 154, 155, 133, 243],
+"EYELINER_RIGHT": [463, 362, 382, 381, 380, 374, 373, 390, 249, 263, 359, 446, 255, 339, 254, 253, 252, 256, 341, 463],
+"EYESHADOW_LEFT": [226, 247, 30, 29, 27, 28, 56, 190, 243, 173, 157, 158, 159, 160, 161, 246, 33, 130, 226],
+"EYESHADOW_RIGHT": [463, 414, 286, 258, 257, 259, 260, 467, 446, 359, 263, 466, 388, 387, 386, 385, 384, 398, 362, 463],
+"FACE": [152, 148, 176, 149, 150, 136, 172, 58, 132, 93, 234, 127, 162, 21, 54, 103, 67, 109, 10, 338, 297, 332, 284, 251, 389, 454, 323, 401, 361, 435, 288, 397, 365, 379, 378, 400, 377, 152],
+"LIP_UPPER": [61, 185, 40, 39, 37, 0, 267, 269, 270, 409, 291, 308, 415, 310, 312, 13, 82, 81, 80, 191, 78],
+"LIP_LOWER": [61, 146, 91, 181, 84, 17, 314, 405, 321, 375, 291, 308, 324, 402, 317, 14, 87, 178, 88, 95, 78, 61],
+"EYEBROW_LEFT": [55, 107, 66, 105, 63, 70, 46, 53, 52, 65, 55],
+"EYEBROW_RIGHT": [285, 336, 296, 334, 293, 300, 276, 283, 295, 285]
+}
+
+# initialize mediapipe face landmarker (new Tasks API)
+# requires face_landmarker.task model file in the project directory
+_base_options = python.BaseOptions(model_asset_path='face_landmarker.task')
+_options = vision.FaceLandmarkerOptions(
+    base_options=_base_options,
+    num_faces=1,
+)
+_face_landmarker = vision.FaceLandmarker.create_from_options(_options)
+
+
+def _normalized_to_pixel_coordinates(normalized_x, normalized_y, image_width, image_height):
+    x_px = min(math.floor(normalized_x * image_width), image_width - 1)
+    y_px = min(math.floor(normalized_y * image_height), image_height - 1)
+    return (x_px, y_px)
+
+
+# to display image in cv2 window
+def show_image(image: np.array, msg: str = "Loaded Image"):
+    """
+    image : image as np array
+    msg : cv2 window name
+    """
+    image_copy = image.copy()
+    cv2.imshow(msg, image_copy)
+    cv2.waitKey(0)
+    cv2.destroyAllWindows()
+
+
+def read_landmarks(image: np.array):
+    """
+    image : image as np.array
+    """
+    landmark_cordinates = {}
+    # convert BGR to RGB, then wrap in mediapipe Image
+    rgb_image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
+    mp_image = mp.Image(image_format=mp.ImageFormat.SRGB, data=rgb_image)
+
+    # detect facial landmarks (returns normalized points 0 to 477)
+    results = _face_landmarker.detect(mp_image)
+    face_landmarks = results.face_landmarks[0]
+
+    # convert normalized points w.r.to image dimensions
+    for idx, landmark in enumerate(face_landmarks):
+        landmark_px = _normalized_to_pixel_coordinates(
+            landmark.x, landmark.y, image.shape[1], image.shape[0]
+        )
+        # create a map of facial landmarks to (x,y) coordinates
+        if landmark_px:
+            landmark_cordinates[idx] = landmark_px
+    return landmark_cordinates
+
+
+# based on input facial features create make w.r.to colors
+def add_mask(
+    mask: np.array, idx_to_coordinates: dict, face_connections: list, colors: list
+):
+    """
+    mask: image filled with 0's
+    idx_to_coordinates : dict with (x,y) cordinates for each face landmarks
+    face_connections : list of (x,y) cordinates for each facial features
+    colors : list of [B,G,R] color for each features
+    """
+    for i, connection in enumerate(face_connections):
+        # extract (x,y) w.r.to image for each cordinates
+        points = np.array([idx_to_coordinates[idx] for idx in connection])
+        # make a shape of feature in the mask and add color
+        cv2.fillPoly(mask, [points], colors[i])
+
+    # smoothening of image
+    mask = cv2.GaussianBlur(mask, (7, 7), 4)
+    return mask

webcam.py

@@ -0,0 +1,64 @@
+import cv2
+import numpy as np
+import pyvirtualcam
+from utils import read_landmarks, add_mask, face_points
+
+face_elements = ["LIP_LOWER", "LIP_UPPER", "EYEBROW_LEFT", "EYEBROW_RIGHT",
+                 "EYELINER_LEFT", "EYELINER_RIGHT", "EYESHADOW_LEFT", "EYESHADOW_RIGHT"]
+colors_map = {
+    # upper lip and lower lips
+    "LIP_UPPER": [0, 0, 255],  # Red in BGR
+    "LIP_LOWER": [0, 0, 255],  # Red in BGR
+    # eyeliner
+    "EYELINER_LEFT": [139, 0, 0],  # Dark Blue in BGR
+    "EYELINER_RIGHT": [139, 0, 0],  # Dark Blue in BGR
+    # eye shadow
+    "EYESHADOW_LEFT": [0, 100, 0],  # Dark Green in BGR
+    "EYESHADOW_RIGHT": [0, 100, 0],  # Dark Green in BGR
+    # eye brow
+    "EYEBROW_LEFT": [19, 69, 139],  # Dark Brown in BGR
+    "EYEBROW_RIGHT": [19, 69, 139],  # Dark Brown in BGR
+}
+
+face_connections = [face_points[idx] for idx in face_elements]
+colors = [colors_map[idx] for idx in face_elements]
+
+cap = cv2.VideoCapture(0)
+if not cap.isOpened():
+    raise RuntimeError("Could not open webcam. Make sure Zoom is not already using it.")
+
+width  = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
+height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
+fps    = cap.get(cv2.CAP_PROP_FPS) or 30
+
+print(f"Webcam opened: {width}x{height} @ {fps}fps")
+
+with pyvirtualcam.Camera(width=width, height=height, fps=fps) as cam:
+    print(f"Virtual camera running: {cam.device}")
+    print("Press Q in the preview window to quit.")
+    while True:
+        ret, frame = cap.read()
+        if not ret:
+            continue  # skip bad frames, don't exit
+        # frame = cv2.flip(frame, 1)  # mirror horizontally
+
+        try:
+            landmark_coordinates = read_landmarks(frame)
+            mask = np.zeros_like(frame)
+            mask = add_mask(mask, idx_to_coordinates=landmark_coordinates,
+                            face_connections=face_connections, colors=colors)
+            output = cv2.addWeighted(frame, 1.0, mask, 0.2, 1.0)
+        except (IndexError, KeyError):
+            output = frame  # no face detected, pass through plain video
+
+        # show local preview so you can confirm filter is working
+        cv2.imshow("Virtual Makeup Preview (Q to quit)", output)
+        if cv2.waitKey(1) & 0xFF == ord('q'):
+            break
+
+        # pyvirtualcam expects RGB
+        cam.send(cv2.cvtColor(output, cv2.COLOR_BGR2RGB))
+        cam.sleep_until_next_frame()
+
+cap.release()
+cv2.destroyAllWindows()