Face Recognition using OpenCV and Machine Learning

Face recognition in Video Output

Data Collection

The data can be collected using the video-cam, or from an image already available.

Working

Working Steps

• OpenCV library along with haarcascade file used for detecting frontal face is used to detect faces in the image/video. The portion of image containing the image is cropped, resized and converted to gray-scale(1D).

• The array output of the gray-scale image along with the name (which is our label/dependent variable) that the user inputs when the code is executed is our training data. This data is stored as a .npy file.

• The model is trained using the SVM classifier.

How To Use

1. Clone the project
2. Navigate to the src folder
3. If you have an image input, run the fetch_face_data(img).py file.
4. If you want to take input from webcam, run the fetch_face_data(vid).py file and also press c on your keyboard after the face is detected to collect the numpy data. Then exit by pressing q.
5. Run the face_recog.py file.

Output

The model was trained on two different images for the following output.