This project aims to recognize faces using OpenCV and the LBPH (Local Binary Patterns Histograms) face recognizer. The code detects faces in images, trains a model using those faces, and then uses the model to recognize faces in new images. Additionally, it includes functionality for recognizing eyes using a separate Haar Cascade classifier.
To run this project, you'll need to install the following dependencies:
- OpenCV
- NumPy
You can install these packages using pip:
pip install numpy opencv-python- Upload your images in the required structure.
- Ensure that the images are organized in folders named after the person in the images.
- Detect faces in images and create a training set:
create_training_set()
print(f"Length of features: {len(features)}")
print(f"Length of labels: {len(labels)}")- Train the LBPH face recognizer:
face_recognizer = cv.face.LBPHFaceRecognizer_create()
np_features = np.array(features, dtype='object')
np_labels = np.array(labels)
face_recognizer.train(np_features, np_labels)
np.save('features.npy', np_features)
np.save('labels.npy', np_labels)
face_recognizer.save('faces_trained.yml')- Recognize a single image:
def recognize_a_single_image(image_path):
image = cv.imread(image_path)
gray = cv.cvtColor(image,cv.COLOR_BGR2GRAY)
faces_rect = haar_cascade.detectMultiScale(gray, scaleFactor=1.1, minNeighbors=4)
for(x,y,w,h) in faces_rect:
face_region_of_interest = gray[y:y+h, x:x+w]
label, accuracy = facesRecognizer.predict(face_region_of_interest)
print(f"This is {people[label]} with an accuracy of {accuracy}%")- Recognize faces in a folder of images:
def recognize_a_folder_of_images(folder_path):
for images in os.listdir(folder_path):
image_path = os.path.join(folder_path + "/", images)
if check_extension(image_path) == 1:
image_read = cv.imread(image_path)
grayed = cv.cvtColor(image_read, cv.COLOR_BGR2GRAY)
faces_rect = haar_cascade.detectMultiScale(grayed, 1.1, 4)
for(x,y,w,h) in faces_rect:
face_region_of_interest = grayed[y:y+h, x:x+w]
label, accuracy = facesRecognizer.predict(face_region_of_interest)
print(f"This is {people[label]} with an accuracy of {accuracy}%")- Detect eyes in images and create a training set:
def create_eye_training_set():
for person in people:
path = DIR + person
label = people.index(person)
for images in os.listdir(path):
image_path = os.path.join(path, images)
if check_extension(image_path) == 1:
image = cv.imread(image_path)
grayed = cv.cvtColor(image, cv.COLOR_BGR2GRAY)
eye_rect = haar_eye_cascade.detectMultiScale(image, 1.1, 4)
for(x,y,w,h) in eye_rect:
region_of_interest = image[y:y+h, x:x+w]
eye_features.append(region_of_interest)
eye_labels.append(label)
else:
continue
create_eye_training_set()
print(f'Length of features: {len(eye_features)}')
print(f'Length of labels: {len(eye_labels)}')harr_face_default.xml: Haar Cascade classifier for face detection.features.npy,labels.npy: Numpy arrays storing the training features and labels.faces_trained.yml: Trained LBPH face recognizer model.haar_eye.xml: Haar Cascade classifier for eye detection./content/persons/: Directory containing training images organized in subdirectories named after each person.
This project is licensed under the MIT License. See the LICENSE file for details.
This documentation provides an overview of the project's purpose, installation steps, usage instructions, file structure, and license information to help users understand and utilize the code effectively.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent