import cv2 import numpy as np import matplotlib.pyplot as plt from matplotlib.collections import PolyCollection from scipy.spatial import Delaunay import os import urllib.request import tempfile import shutil # Get the directory where this script is located script_dir = os.path.dirname(os.path.abspath(__file__)) # Download the face landmarker model if not present model_path_src = os.path.join(script_dir, "face_landmarker.task") if not os.path.exists(model_path_src): print("Downloading MediaPipe Face Landmarker model...") model_url = "https://storage.googleapis.com/mediapipe-models/face_landmarker/face_landmarker/float16/1/face_landmarker.task" urllib.request.urlretrieve(model_url, model_path_src) print("Model downloaded successfully!") # Copy to temp dir (MediaPipe cannot handle Unicode in paths) model_path = shutil.copy2(model_path_src, tempfile.mkdtemp()) # Import MediaPipe import mediapipe as mp from mediapipe.tasks import python from mediapipe.tasks.python import vision def detect_face_landmarks(image_rgb, model_path): """ Detect facial landmarks using MediaPipe Face Landmarker. Parameters: image_rgb: RGB image as numpy array model_path: Path to the face_landmarker.task model file Returns: List of (x, y) tuples for each landmark, or None if no face detected """ # Create MediaPipe Image mp_image = mp.Image(image_format=mp.ImageFormat.SRGB, data=image_rgb) # Configure and create detector base_options = python.BaseOptions(model_asset_path=model_path) options = vision.FaceLandmarkerOptions( base_options=base_options, num_faces=1 ) detector = vision.FaceLandmarker.create_from_options(options) # Detect landmarks result = detector.detect(mp_image) detector.close() if not result.face_landmarks: return None # Extract landmark coordinates as pixel positions height, width = image_rgb.shape[:2] landmarks = [] for lm in result.face_landmarks[0]: x = lm.x * width y = lm.y * height landmarks.append((x, y)) return landmarks def create_lowpoly_face(image_rgb, landmarks, add_boundary=True): """ Create a low-poly rendering of a face using Delaunay triangulation. Parameters: image_rgb: Original RGB image for color sampling landmarks: List of (x, y) landmark coordinates add_boundary: If True, add corner points for full coverage Returns: triangles: List of triangle vertex arrays colors: List of RGB colors for each triangle triangulation: The scipy Delaunay triangulation object """ height, width = image_rgb.shape[:2] # Convert landmarks to numpy array points = np.array(landmarks) # Optionally add boundary points for better coverage if add_boundary: # Add corner points corners = np.array([ [0, 0], [width, 0], [width, height], [0, height] ]) # Add edge midpoints for smoother boundary edges = np.array([ [width/2, 0], [width, height/2], [width/2, height], [0, height/2] ]) points = np.vstack([points, corners, edges]) # Compute Delaunay triangulation # This connects all points into non-overlapping triangles triangulation = Delaunay(points) # Build colored triangles triangles = [] colors = [] for simplex in triangulation.simplices: # Get the three vertices of this triangle triangle = points[simplex] triangles.append(triangle) # Calculate centroid for color sampling centroid = np.mean(triangle, axis=0) cx = int(np.clip(centroid[0], 0, width - 1)) cy = int(np.clip(centroid[1], 0, height - 1)) # Sample color from original image at centroid # Note: image is [row, col] = [y, x] color = image_rgb[cy, cx] / 255.0 colors.append(color) return triangles, colors, triangulation def render_lowpoly(triangles, colors, width, height, show_edges=False): """ Render the low-poly triangles using matplotlib. Parameters: triangles: List of triangle vertex arrays colors: List of RGB colors for each triangle width, height: Image dimensions show_edges: If True, draw triangle edges Returns: Rendered figure """ fig, ax = plt.subplots(figsize=(10, 10)) # Create polygon collection for efficient rendering edge_color = 'white' if show_edges else 'none' line_width = 0.5 if show_edges else 0 collection = PolyCollection( triangles, facecolors=colors, edgecolors=edge_color, linewidths=line_width ) ax.add_collection(collection) ax.set_xlim(0, width) ax.set_ylim(height, 0) # Flip Y-axis to match image coordinates ax.set_aspect('equal') ax.axis('off') return fig # Main execution if __name__ == '__main__': # Load the sample face image image_path = os.path.join(script_dir, "sample_face.jpg") image_bgr = cv2.imread(image_path) if image_bgr is None: print(f"Error: Could not load image from {image_path}") exit(1) # Convert BGR to RGB image_rgb = cv2.cvtColor(image_bgr, cv2.COLOR_BGR2RGB) height, width = image_rgb.shape[:2] print(f"Image size: {width} x {height}") print("Detecting facial landmarks...") # Detect landmarks landmarks = detect_face_landmarks(image_rgb, model_path) if landmarks is None: print("No face detected! Please use an image with a visible face.") exit(1) print(f"Detected {len(landmarks)} facial landmarks") # Create low-poly version print("Creating low-poly triangulation...") triangles, colors, triangulation = create_lowpoly_face( image_rgb, landmarks, add_boundary=True ) print(f"Created {len(triangles)} triangles") # Render and save the low-poly result print("Rendering low-poly face art...") fig = render_lowpoly(triangles, colors, width, height, show_edges=False) output_path = os.path.join(script_dir, "lowpoly_face_output.png") fig.savefig(output_path, dpi=150, bbox_inches='tight', pad_inches=0) plt.close(fig) print(f"Low-poly image saved to: {output_path}") # Create comparison grid (original vs low-poly) print("Creating comparison grid...") fig, axes = plt.subplots(1, 2, figsize=(16, 8)) # Original image axes[0].imshow(image_rgb) axes[0].set_title('Original Image', fontsize=16, fontweight='bold') axes[0].axis('off') # Low-poly version collection = PolyCollection( triangles, facecolors=colors, edgecolors='none' ) axes[1].add_collection(collection) axes[1].set_xlim(0, width) axes[1].set_ylim(height, 0) axes[1].set_aspect('equal') axes[1].set_title('Low-Poly Face Art', fontsize=16, fontweight='bold') axes[1].axis('off') plt.tight_layout() comparison_path = os.path.join(script_dir, "comparison_grid.png") fig.savefig(comparison_path, dpi=150, bbox_inches='tight') plt.close(fig) print(f"Comparison grid saved to: {comparison_path}") # Also create a version with edges for artistic effect fig = render_lowpoly(triangles, colors, width, height, show_edges=True) edges_path = os.path.join(script_dir, "lowpoly_face_edges.png") fig.savefig(edges_path, dpi=150, bbox_inches='tight', pad_inches=0) plt.close(fig) print(f"Low-poly with edges saved to: {edges_path}") print("\nDone! Check the generated images.")