OpenCV 16: Face Recognition

I used OpenCV’s built-in face recognizer and trained it using all of the images I downloaded from Google in these two folders.

There are three modules we need to import – os, cv2 and numpy.

import os
import cv2 as cv
import numpy as np

1. Create a list of all the animals in the folders, in my case chicken and tiger.
2. Create a variable and set it equal to the base folder that contains the 2 folders of the image of the animals.
3. Create two lists for storing the training set called features and labels. The features are the image array of the animals whereas the second list is the corresponding labels.
4. Create a function to set up the training set for the model.

animals = ['Chicken','Tiger']
DIR = r'/Users/richzha/Desktop/CV/Face_recognition2'

features = []
labels = []

def create_train():

In the Function:

1. First loop over all the animals in the list.
2. Grab the path for that particular animal by joining the path of the base folder to the name of the animal.
3. Create a label for the images e.g. tiger

    for animal in animals:
        path = os.path.join(DIR, animal)
        label = animals.index(animal)

Now that we are inside each folder, we are going to loop through all the images in that folder.

1. Grabe the path for that particular image in the folder of the animal.
2. Read in that image from the path.
3. Convert the image to a gray scale
4. Append the gray image to the features list and append the label of the animal to the labels list.

for img in os.listdir(path):
    img_path = os.path.join(path, img)
    if img_path.endswith("jpeg") or img_path.endswith("jpg"):

        img_array = cv.imread(img_path)
        gray = cv.cvtColor(img_array, cv.COLOR_BGR2GRAY)

        features.append(gray)
        labels.append(label)

Now the training function is complete and it could be called on the main code.

create_train()

After that, we can use the features and labels list to train our recognizer.

1. Convert the features and labels list to numpy arrays.
2. Instantiate our face recognizer as an instance of cv.face.LBPHFaceRecognizer_create() class.
3. Train the recognizer on the features list and labels list.
4. Save the features and labels list for later use.
5. Save the trained model in a .yml source file so that we can use it in another file.

features = np.array(features, dtype=object)
labels = np.array(labels)

face_recognizer = cv.face.LBPHFaceRecognizer_create()

face_recognizer.train(features, labels)

face_recognizer.save('Tiger_chicken_trained.yml')

np.save('Features_tiger_chicken.npy', features)
np.save('Labels_tiger_chicken.npy', labels)

By now, the model for detecting tigers and chickens is successfully trained.

Appling the Trained Model

Create a new file, read in the trained model in the .yml file and create a list of all the animals.

import numpy as np
import cv2 as cv

TC_recognizer = cv.face.LBPHFaceRecognizer_create()
TC_recognizer.read('Tiger_chicken_trained.yml')

people = ['Chicken','Tiger']

1. Select an image of a chicken and read that image in using OpenCV
2. Convert the selected image to a gray-scale image.
3. Predict what animal it is using the face recognizer. Create the label and confidence value variables which is assigned to the output of the trained model.
4. Print the output from the model.
5. Put the label on the image passed to the model and output that image.

img = cv.imread('/Desktop/CV/Face_recognition2/Chicken/Chicken.jpg')
gray = cv.cvtColor(img, cv.COLOR_BGR2GRAY)


label, confidence = TC_recognizer.predict(gray)

print(f'Label = {people[label]} with a confidence of {confidence}')

cv.putText(img, str(people[label]), (20,20), cv.FONT_HERSHEY_COMPLEX, 1.0, (0,0,0), thickness=2)

cv.imshow('Detected Image', img)

cv.waitKey(0)

Test with Images