Face recognition is one of the techniques in the fields of image analysis and computer vision due to its wide practical applications in biometrics, information security, law enforcement and surveillance systems. Factors such as illumination, emotion, occlusion, facial expressions and poses, which greatly affect the performance in achieving efficient and robust face recognition systems. Recent development in deep learning and neural networks have made it possible to achieve promising results in numerous fields including pattern recognition and image processing.
There has been great advancement in face recognition, starting with the pioneer work for detecting frontal-face in real time along with low computational complexity. Other than this most of the initial approaches in face recognition used up-right images without many variations in pose, illumination, occlusion.
Deep learning is one of the most advanced forms of machine learning that uses multiple layers to extract features from a data set and learns in a supervised as well as unsupervised manner. It involves different models based on the neural networks and it has improved the ability of classification, recognition, detection, and localization.
Face recognition may be divided into two steps, which are listed below:
Step 1: Identifying
Individuals can be identified by finding their faces in each photograph. The initial stage of a facial recognition system the identifying step guarantees that the algorithm recognizes the image as a face image. This information is then used to identify the faces in the image. The identification stage compares the face in the image to the other faces to determine the identity of the face in the image. As a result, this is a multiclass classification issue.
Step 2: Confirmation
The verification is concerned with identity confirmation based on a facial picture as input It does a one-to-one match by accepting or rejecting the identity.