face_recognition.py

#face recognition using keras and tensor flow..
#we trained model online and exported code to here


from keras.model import load_model
from PIL import Image, ImageOps
import numpy as np
import cv2
from email.mime import image

cap = cv2.VideoCapture(0)
# Load the model
model = load_model('keras_model.h5')

#load labels
with open('labels.txt','r') as f:
  class_names = f.read().split('\n') 

  frame_width = int(cap.get(3))
  frame_height = int(cap.get(4))

  storage_size=(frame_width, frame_height)
  result=cv2.VideoWriter('', cv2.VideoWriter_fourcc(*'MJPG'), 10, storage_size)

while True:
    _, image = cap.read()
    # Create the array of the right shape to feed into the keras model
# The 'length' or number of images you can put into the array is
# determined by the first position in the shape tuple, in this case 1.
    data = np.ndarray(shape=(1, 224, 224, 3), dtype=np.float32)
# Replace this with the path to your image
    
#resize the image to a 224x224 with the same strategy as in TM2:
#resizing the image to be at least 224x224 and then cropping from the center
    size = (224, 224)
    image_PIL = Image.fromarray(image)
    image = ImageOps.fit(image_PIL, size, Image.Resampling.LANCZOS)

#turn the image into a numpy array
    image_array = np.asarray(image)
# Normalize the image
    normalized_image_array = (image_array.astype(np.float32) / 127.0) - 1
# Load the image into the array
    data[0] = normalized_image_array

# run the inference
    prediction = model.predict(data)
    print(prediction)
    index = np.argmax(prediction)
    class_name = class_names[index]
    confidence_score =prediction

    print("class:", class_name)
    print("confidence score:", confidence_score)
   
    cv2.putText(image,str(class_name),(10,70),cv2.FONT_HERSHEY_COMPLEX,1,(0,255,0), None)
    cv2.imshow("nestor", image)
    key=cv2.waitKey(1)
    if key == ord('q'):
        break
image.release()
result.release()
cv2.destroyAllWindows