• 資訊安全概論

程式安全-人臉辨識

• 講義

• 人臉辨識

  原始圖片

  

  • 實作1-顯示圖片

  import cv2
  from google.colab.patches import cv2_imshow # Import cv2_imshow from google.colab.patches
  
  img = cv2.imread('face.png')   # 開啟圖片，預設使用 cv2.IMREAD_COLOR 模式
  cv2_imshow(img)                # Use cv2_imshow instead of cv2.imshow to display the image
  cv2.waitKey(0)                 # 按下任意鍵停止
  cv2.destroyAllWindows()        # 結束所有圖片視窗

  執行結果

  

  • 實作2-顯示灰階圖片

  import cv2
  from google.colab.patches import cv2_imshow         # Import cv2_imshow for Colab
  
  img = cv2.imread('face.png', cv2.IMREAD_GRAYSCALE)  # 使用 cv2.IMREAD_GRAYSCALE 模式
  cv2_imshow(img)                                     # Use cv2_imshow instead of cv2.imshow
  cv2.waitKey(2000)                                   # 等待兩秒 ( 2000 毫秒 ) 後關閉圖片視窗
  cv2.destroyAllWindows()

  執行結果

  

  • 實作3-人臉辨識

  文件  haarcascade_frontalface_default.xml

  import cv2
  from google.colab.patches import cv2_imshow # Import cv2_imshow for Colab display
  
  img = cv2.imread('face.png')
  gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)   # 將圖片轉成灰階
  
  face_cascade = cv2.CascadeClassifier("haarcascade_frontalface_default.xml")   # 載入人臉模型
  faces = face_cascade.detectMultiScale(gray)    # 偵測人臉
  
  for (x, y, w, h) in faces:
      cv2.rectangle(img, (x, y), (x+w, y+h), (0, 255, 0), 2)    # 利用 for 迴圈，抓取每個人臉屬性，繪製方框
  
  cv2_imshow(img) # Use cv2_imshow instead of cv2.imshow for display in Colab
  cv2.waitKey(0) # 按下任意鍵停止
  cv2.destroyAllWindows()

  執行結果

  

  • 實作4-五官辨識

  文件 

  眼睛特徵模型：haarcascade_eye.xml

  嘴巴特徵模型：haarcascade_mcs_mouth.xml

  鼻子特徵模型：haarcascade_mcs_nose.xml

  import cv2
  from google.colab.patches import cv2_imshow # Import cv2_imshow for Colab display
  
  img = cv2.imread('face.png')
  gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)                       # 圖片轉灰階
  #gray = cv2.medianBlur(gray, 5)                                    # 如果一直偵測到雜訊，可使用模糊的方式去除雜訊
  
  eye_cascade = cv2.CascadeClassifier("haarcascade_eye.xml")         # 使用眼睛模型
  eyes = eye_cascade.detectMultiScale(gray)                          # 偵測眼睛
  for (x, y, w, h) in eyes:
      cv2.rectangle(img, (x, y), (x+w, y+h), (0, 255, 0), 2)         # 標記綠色方框
  
  mouth_cascade = cv2.CascadeClassifier("haarcascade_mcs_mouth.xml") # 使用嘴巴模型
  mouths = mouth_cascade.detectMultiScale(gray)                      # 偵測嘴巴
  for (x, y, w, h) in mouths:
      cv2.rectangle(img, (x, y), (x+w, y+h), (0, 0, 255), 2)         # 標記紅色方框
  
  nose_cascade = cv2.CascadeClassifier("haarcascade_mcs_nose.xml")   # 使用鼻子模型
  noses = nose_cascade.detectMultiScale(gray)                        # 偵測鼻子
  for (x, y, w, h) in noses:
      cv2.rectangle(img, (x, y), (x+w, y+h), (255, 0, 0), 2)         # 標記藍色方框
  
  # Replace cv2.imshow with cv2_imshow
  cv2_imshow(img)  # Use cv2_imshow instead of cv2.imshow to display the image in Colab
  cv2.waitKey(0)   # 按下任意鍵停止
  cv2.destroyAllWindows()

  執行結果

  

  • 實作5-資訊分析

   !pip install deepface

  程式

  import cv2
  from deepface import DeepFace
  import numpy as np
  from google.colab.patches import cv2_imshow # Import cv2_imshow for Colab display
  
  
  img = cv2.imread('face.png')
  try:
      emotion = DeepFace.analyze(img, actions=['emotion'])  # 情緒
      age = DeepFace.analyze(img, actions=['age'])          # 年齡
      race = DeepFace.analyze(img, actions=['race'])        # 人種
      gender = DeepFace.analyze(img, actions=['gender'])    # 性別
  
      print(emotion[0]['dominant_emotion'])
      print(age[0]['age'])
      print(race[0]['dominant_race'])
      print(gender[0]['gender'])
  except:
      pass
  
  cv2_imshow(img) # Use cv2_imshow instead of cv2.imshow for display in Colab
  cv2.waitKey(0)
  cv2.destroyAllWindows()

  執行結果

  happy
  28
  white
  {'Woman': 0.0007532069048465928, 'Man': 99.99924898147583}

  實作6-網格分析

  !pip install -q mediapipe

  !wget -O face_landmarker_v2_with_blendshapes.task -q https://storage.googleapis.com/mediapipe-models/face_landmarker/face_landmarker/float16/1/face_landmarker.task

  #@markdown We implemented some functions to visualize the face landmark detection results. <br/> Run the following cell to activate the functions.
  
  from mediapipe import solutions
  from mediapipe.framework.formats import landmark_pb2
  import numpy as np
  import matplotlib.pyplot as plt
  
  
  def draw_landmarks_on_image(rgb_image, detection_result):
    face_landmarks_list = detection_result.face_landmarks
    annotated_image = np.copy(rgb_image)
  
    # Loop through the detected faces to visualize.
    for idx in range(len(face_landmarks_list)):
      face_landmarks = face_landmarks_list[idx]
  
      # Draw the face landmarks.
      face_landmarks_proto = landmark_pb2.NormalizedLandmarkList()
      face_landmarks_proto.landmark.extend([
        landmark_pb2.NormalizedLandmark(x=landmark.x, y=landmark.y, z=landmark.z) for landmark in face_landmarks
      ])
  
      solutions.drawing_utils.draw_landmarks(
          image=annotated_image,
          landmark_list=face_landmarks_proto,
          connections=mp.solutions.face_mesh.FACEMESH_TESSELATION,
          landmark_drawing_spec=None,
          connection_drawing_spec=mp.solutions.drawing_styles
          .get_default_face_mesh_tesselation_style())
      solutions.drawing_utils.draw_landmarks(
          image=annotated_image,
          landmark_list=face_landmarks_proto,
          connections=mp.solutions.face_mesh.FACEMESH_CONTOURS,
          landmark_drawing_spec=None,
          connection_drawing_spec=mp.solutions.drawing_styles
          .get_default_face_mesh_contours_style())
      solutions.drawing_utils.draw_landmarks(
          image=annotated_image,
          landmark_list=face_landmarks_proto,
          connections=mp.solutions.face_mesh.FACEMESH_IRISES,
            landmark_drawing_spec=None,
            connection_drawing_spec=mp.solutions.drawing_styles
            .get_default_face_mesh_iris_connections_style())
  
    return annotated_image
  
  def plot_face_blendshapes_bar_graph(face_blendshapes):
    # Extract the face blendshapes category names and scores.
    face_blendshapes_names = [face_blendshapes_category.category_name for face_blendshapes_category in face_blendshapes]
    face_blendshapes_scores = [face_blendshapes_category.score for face_blendshapes_category in face_blendshapes]
    # The blendshapes are ordered in decreasing score value.
    face_blendshapes_ranks = range(len(face_blendshapes_names))
  
    fig, ax = plt.subplots(figsize=(12, 12))
    bar = ax.barh(face_blendshapes_ranks, face_blendshapes_scores, label=[str(x) for x in face_blendshapes_ranks])
    ax.set_yticks(face_blendshapes_ranks, face_blendshapes_names)
    ax.invert_yaxis()
  
    # Label each bar with values
    for score, patch in zip(face_blendshapes_scores, bar.patches):
      plt.text(patch.get_x() + patch.get_width(), patch.get_y(), f"{score:.4f}", va="top")
  
    ax.set_xlabel('Score')
    ax.set_title("Face Blendshapes")
    plt.tight_layout()
    plt.show()

  import cv2
  from google.colab.patches import cv2_imshow
  
  img = cv2.imread("face.png")
  cv2_imshow(img)

   

  # STEP 1: Import the necessary modules.
  import mediapipe as mp
  from mediapipe.tasks import python
  from mediapipe.tasks.python import vision
  
  # STEP 2: Create an FaceLandmarker object.
  base_options = python.BaseOptions(model_asset_path='face_landmarker_v2_with_blendshapes.task')
  options = vision.FaceLandmarkerOptions(base_options=base_options,
                                         output_face_blendshapes=True,
                                         output_facial_transformation_matrixes=True,
                                         num_faces=1)
  detector = vision.FaceLandmarker.create_from_options(options)
  
  # STEP 3: Load the input image.
  image = mp.Image.create_from_file("face.png")
  
  # STEP 4: Detect face landmarks from the input image.
  detection_result = detector.detect(image)
  
  # STEP 5: Process the detection result. In this case, visualize it.
  annotated_image = draw_landmarks_on_image(image.numpy_view(), detection_result)
  cv2_imshow(cv2.cvtColor(annotated_image, cv2.COLOR_RGB2BGR))

  執行結果

  

  plot_face_blendshapes_bar_graph(detection_result.face_blendshapes[0])

  執行結果

  

   

  print(detection_result.facial_transformation_matrixes)

  執行結果

  

  議題討論-人臉辨識的隱憂

  文章來源：https://ithelp.ithome.com.tw/articles/10276505

  在人臉辨識的興盛下，如同科技進步一般，水能載舟，亦能覆舟，有如此多的優點也會也缺點，如先前講的隱私權及要如何確保每次辨識下都是正確的，換句話說就是辨識時的不穩定。

  1.隱私權：
  在模型進行訓練時就需要大量資料來進行訓練，所以需要很多人臉的資料，在資料的收集就十分困難，目前就有使用收集意修資料後，用這些資料來產生訓練資料，就是生出假的人臉，使用先前收集的資料學習組合成一張以假亂真的臉，或是使用資料增量的方式來增加訓練資料，如不同的光影變化，或是使用3D建模的方式來擷取人臉的不同角度，讓可以偵測的角度變大。

  2.辨識時的不穩定：
  使用更加完善的模型架構，如前面所提到的各種人臉常用模型，或是使用更強健的演算法來降低辨識不穩定的發生率，因為人臉的變化有光線、角度、年紀、化妝等可以改變人臉外表的因素，維持辨識的穩定，可能需要在辨識時持續更新或增加資料庫的影像。

  在未來人臉辨識系統發展，可以有更強大的辨識系統來突破這些限制，使人類的生活更加便利。