This article mainly introduces “ How to use it? Python The code realizes face recognition ” Knowledge about , Xiaobian shows you the operation process through practical cases , The operation method is simple and fast , Practical , Hope this article “ How to use it? Python The code realizes face recognition ” The article can help you solve problems .
Environmental requirements :
Ubuntu17.10
Python 2.7.14
Environment building :
1. install Ubuntu17.10 > Installation steps are here
2. install Python2.7.14 (Ubuntu17.10 Default Python Version is 2.7.14)
3. install git 、cmake 、 python-pip
# install git $ sudo apt-get install -y git # install cmake $ sudo apt-get install -y cmake # install python-pip $ sudo apt-get install -y python-pip
4. Installation and compilation dlib
install face_recognition You need to install and compile before this dlib
# compile dlib Install before boost $ sudo apt-get install libboost-all-dev # Start compilation dlib # clone dlib Source code $ git clone https://github.com/davisking/dlib.git $ cd dlib $ mkdir build $ cd build $ cmake .. -DDLIB_USE_CUDA=0 -DUSE_AVX_INSTRUCTIONS=1 $ cmake --build .( Notice that there is a space in the middle ) $ cd .. $ python setup.py install --yes USE_AVX_INSTRUCTIONS --no DLIB_USE_CUDA
5. install face_recognition
# install face_recognition $ pip install face_recognition # install face_recognition It will be installed automatically numpy、scipy etc.
After the environment is built , Input... At the terminal face_recognition Command to see if it was successful
1. First you need to provide a folder , Inside are pictures of all the people you want the system to know . Each of them has a picture , Pictures are named after people :
known_people There is... Under the folder babe、 Jackie Chan 、 Joey Yung's photo
2. Next , You need to prepare another folder , Inside is the picture you want to identify : unknown_pic Under the folder is the image to be recognized , Among them, Han Hong is unknown to the machine
3. Then you can run face_recognition The command , Pass in the two folders just prepared as parameters , The command will return who appears in the image that needs to be recognized :
Recognition success !!!
# filename : find_faces_in_picture.py # -*- coding: utf-8 -*- # Import pil modular , Install with commands apt-get install python-Imaging from PIL import Image # Import face_recogntion modular , Install with commands pip install face_recognition import face_recognition # take jpg File loading to numpy Array image = face_recognition.load_image_file("/opt/face/unknown_pic/all_star.jpg") # Use the default giving HOG The model finds all faces in the image # This method is already quite accurate , But not as good as CNN The model is so accurate , Because I didn't use GPU Speed up # See also : find_faces_in_picture_cnn.py face_locations = face_recognition.face_locations(image) # Use CNN Model # face_locations = face_recognition.face_locations(image, number_of_times_to_upsample=0, model="cnn") # Print : I found it in the picture How many? Zhang Renren print("I found {} face(s) in this photograph.".format(len(face_locations))) # Loop through all the faces found for face_location in face_locations: # Print the position information of each face top, right, bottom, left = face_location print("A face is located at pixel location Top: {}, Left: {}, Bottom: {}, Right: {}".format(top, left, bottom, right)) # Specify the location information of the face , Then display the face picture face_image = image[top:bottom, left:right] pil_image = Image.fromarray(face_image) pil_image.show()
The following figure shows the image used for recognition
# perform python file $ python find_faces_in_picture.py
Recognize from the picture 7 Zhang Renren , And show it , Here's the picture
# filename : find_facial_features_in_picture.py # -*- coding: utf-8 -*- # Import pil modular , Install with commands apt-get install python-Imaging from PIL import Image, ImageDraw # Import face_recogntion modular , Install with commands pip install face_recognition import face_recognition # take jpg File loading to numpy Array image = face_recognition.load_image_file("biden.jpg") # Find all facial features of all faces in the image face_landmarks_list = face_recognition.face_landmarks(image) print("I found {} face(s) in this photograph.".format(len(face_landmarks_list))) for face_landmarks in face_landmarks_list: # Print the location of each facial feature in this image facial_features = [ 'chin', 'left_eyebrow', 'right_eyebrow', 'nose_bridge', 'nose_tip', 'left_eye', 'right_eye', 'top_lip', 'bottom_lip' ] for facial_feature in facial_features: print("The {} in this face has the following points: {}".format(facial_feature, face_landmarks[facial_feature])) # Let's depict each facial feature in the image ! pil_image = Image.fromarray(image) d = ImageDraw.Draw(pil_image) for facial_feature in facial_features: d.line(face_landmarks[facial_feature], width=5) pil_image.show()
Automatically recognize facial features ( outline )
# filename : recognize_faces_in_pictures.py # -*- conding: utf-8 -*- # Import face_recogntion modular , Install with commands pip install face_recognition import face_recognition # take jpg File loading to numpy Array babe_image = face_recognition.load_image_file("/opt/face/known_people/babe.jpeg") Rong_zhu_er_image = face_recognition.load_image_file("/opt/face/known_people/Rong zhu er.jpg") unknown_image = face_recognition.load_image_file("/opt/face/unknown_pic/babe2.jpg") # Get the face code of each face in each image file # Because there may be multiple faces in each image , So return a list of codes . # But because I know that each image has only one face , I only care about the first code in each image , So I take the index 0. babe_face_encoding = face_recognition.face_encodings(babe_image)[0] Rong_zhu_er_face_encoding = face_recognition.face_encodings(Rong_zhu_er_image)[0] unknown_face_encoding = face_recognition.face_encodings(unknown_image)[0] known_faces = [ babe_face_encoding, Rong_zhu_er_face_encoding ] # The result is True/false Array of , Unknown face known_faces The result of matching anyone in the array results = face_recognition.compare_faces(known_faces, unknown_face_encoding) print(" This unknown face is Babe Do you ? {}".format(results[0])) print(" This unknown face is Joey Yung Do you ? {}".format(results[1])) print(" This unknown face is New faces we've never seen ? {}".format(not True in results))
The display results are shown in the figure
# filename : digital_makeup.py # -*- coding: utf-8 -*- # Import pil modular , Install with commands apt-get install python-Imaging from PIL import Image, ImageDraw # Import face_recogntion modular , Install with commands pip install face_recognition import face_recognition # take jpg File loading to numpy Array image = face_recognition.load_image_file("biden.jpg") # Find all facial features of all faces in the image face_landmarks_list = face_recognition.face_landmarks(image) for face_landmarks in face_landmarks_list: pil_image = Image.fromarray(image) d = ImageDraw.Draw(pil_image, 'RGBA') # Let eyebrows become a nightmare d.polygon(face_landmarks['left_eyebrow'], fill=(68, 54, 39, 128)) d.polygon(face_landmarks['right_eyebrow'], fill=(68, 54, 39, 128)) d.line(face_landmarks['left_eyebrow'], fill=(68, 54, 39, 150), width=5) d.line(face_landmarks['right_eyebrow'], fill=(68, 54, 39, 150), width=5) # Glossy lips d.polygon(face_landmarks['top_lip'], fill=(150, 0, 0, 128)) d.polygon(face_landmarks['bottom_lip'], fill=(150, 0, 0, 128)) d.line(face_landmarks['top_lip'], fill=(150, 0, 0, 64), width=8) d.line(face_landmarks['bottom_lip'], fill=(150, 0, 0, 64), width=8) # Shining eyes d.polygon(face_landmarks['left_eye'], fill=(255, 255, 255, 30)) d.polygon(face_landmarks['right_eye'], fill=(255, 255, 255, 30)) # Apply some Eyeliner d.line(face_landmarks['left_eye'] + [face_landmarks['left_eye'][0]], fill=(0, 0, 0, 110), width=6) d.line(face_landmarks['right_eye'] + [face_landmarks['right_eye'][0]], fill=(0, 0, 0, 110), width=6) pil_image.show()
The comparison before and after beauty is shown in the following figure :
About “ How to use it? Python The code realizes face recognition ” That's all for , Thanks for reading . If you want to know more about the industry , You can pay attention to the Yisu cloud industry information channel , Xiaobian will update different knowledge points for you every day .