Face Detcetion And Security Alert

# face-detcetion-and-security-alert LiveCam Face Recognition with Multi person matching and Threshold Alert import os import dlib import face_recognition import mediapipe as mp import numpy as np import cv2 from FaceMetrics import FaceMetrics # Initialize Mediapipe Face Mesh mp_face_mesh = mp.solutions.face_mesh mp_drawing = mp.solutions.drawing_utils mp_drawing_styles = mp.solutions.drawing_styles # Set up video capture cap = cv2.VideoCapture(0) EYE_AR_THRESH = 0.25 MOUTH_AR_THRESH = 0.98 EYE_AR_CONSEC_imageS = 3 COUNTER = 5 KNOWN_FACES_DIR = "../registered_faces" # training data TOLERANCE = 0.4 image_THICKNESS = 1 # rectangle thickness FONT_THICKNESS = 2 # font thickness MODEL = "cnn" # convolutional print("loading known faces") known_faces = [] # store known faces here known_names = [] # store name of them here print(dlib.DLIB_USE_CUDA) # ture, if cuda is enabled. ### TRAIN THE CNN MODEL ON KNOWN FACES ############################# for name in os.listdir(KNOWN_FACES_DIR): for filename in os.listdir(f"{KNOWN_FACES_DIR}/{name}"): image = face_recognition.load_image_file(f"{KNOWN_FACES_DIR}/{name}/{filename}") encoding = face_recognition.face_encodings(image)[0] known_faces.append(encoding) known_names.append(name) with mp_face_mesh.FaceMesh(static_image_mode=False, max_num_faces=1, refine_landmarks=True, min_detection_confidence=0.5, min_tracking_confidence=0.5) as face_mesh: while cap.isOpened(): ret, frame = cap.read() if not ret: break # Convert frame to RGB image = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB) image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB) locations = face_recognition.face_locations(image,model=MODEL) # find face encodings = face_recognition.face_encodings(image, locations) # extrac face features for face_encoding, face_location in zip(encodings, locations): if len(face_encoding) == 0: continue # Skip if encoding not found results = face_recognition.compare_faces(known_faces, face_encoding) face_dist = face_recognition.face_distance(known_faces, face_encoding) print(f"Face distances: {face_dist}") matchIndex = np.argmin(face_dist) match = "Unknown" if results[matchIndex] and face_dist[matchIndex]<=TOLERANCE: # if there is a known face match = known_names[matchIndex] # who that person is print(f"Match found: {match}") ### DRAW RECTANGLE AND ##################################### top_left = (face_location[3], face_location[0]) bottom_right = (face_location[1], face_location[2]) color = [0, 255, 0] # green cv2.rectangle(image, top_left, bottom_right, color, image_THICKNESS) top_left = (face_location[3], face_location[2]) bottom_right = (face_location[1], face_location[2]+22) cv2.rectangle(image, top_left, bottom_right, color, cv2.FILLED) cv2.putText(image, match, (face_location[3]+10, face_location[2]+15), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (200, 200, 200), FONT_THICKNESS) image.flags.writeable = False results = face_mesh.process(image) # Convert back to BGR for display image.flags.writeable = True image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR) if results.multi_face_landmarks: for face_landmarks in results.multi_face_landmarks: # Get image dimensions h, w, _ = image.shape # Create FaceMetrics instance landmarks = face_landmarks.landmark # Draw face mesh mp_drawing.draw_landmarks( image=image, landmark_list=face_landmarks, connections=mp_face_mesh.FACEMESH_TESSELATION, landmark_drawing_spec=None, connection_drawing_spec=mp_drawing_styles.get_default_face_mesh_tesselation_style()) mp_drawing.draw_landmarks( image=image, landmark_list=face_landmarks, connections=mp_face_mesh.FACEMESH_CONTOURS, landmark_drawing_spec=None, connection_drawing_spec=mp_drawing_styles.get_default_face_mesh_contours_style()) metrics = FaceMetrics(landmarks, w, h) analysis = metrics.analyze() ear = analysis['common_ear'] mar = analysis['mouth_aspect_ratio'] if ear < EYE_AR_THRESH: COUNTER += 1 if COUNTER >= EYE_AR_CONSEC_imageS: cv2.putText(image, "Eyes Closed!", (500, 20), cv2.FONT_HERSHEY_SIMPLEX, 0.7, (0, 0, 255), 2) else: COUNTER = 0 # MAR Calculation cv2.putText(image, "MAR: {:.2f}".format(mar), (20, 20), cv2.FONT_HERSHEY_SIMPLEX, 0.7, (0, 0, 255), 2) if mar < MOUTH_AR_THRESH: cv2.putText(image, "Yawning!", (500, 20), cv2.FONT_HERSHEY_SIMPLEX, 0.7, (0, 0, 255), 2) # Display analysis results cv2.putText(image, f"EAR: {analysis['common_ear']:.2f}", (10, 60), cv2.FONT_HERSHEY_SIMPLEX, 0.7, (255, 0, 0), 2) cv2.putText(image, f"MAR: {analysis['mouth_aspect_ratio']:.2f}", (10, 90), cv2.FONT_HERSHEY_SIMPLEX, 0.7, (255, 0, 0), 2) cv2.putText(image, f"Direction: {analysis['head_direction']}", (10, 120), cv2.FONT_HERSHEY_SIMPLEX, 0.7, (255, 0, 0), 2) # Show the frame cv2.imshow("Face Metrics", image) # Exit on pressing 'q' if cv2.waitKey(5) & 0xFF == ord('q'): break cap.release() cv2.destroyAllWindows()