Haar Cascade and Convolutional Neural Network Face Detection in Client-Side for Cloud Computing Face Recognition

Abstract

Faces are often used as authentication in security systems, such as facial recognition in running programs or facial recognition for room entry. One important reason is that the human face consists of structures and characteristics that differ from one person to another. In the authentication system, the face will be recognized as a person who has permission access. Many methods have been created and developed to perform image face detection. Two popular methods used today are Haar Cascade and Convolutional Neural Network (CNN). This paper proposed the usage of Haar Cascade and CNN for face detection. Haar Cascade is an algorithm that is used to detect a face quickly and in real-time. At the same time, CNN utilizes the convolution process by moving a convolution (filter) kernel of a specific size to the next image from the result of multiplying the image with the filter used. The results obtained indicate that the system has succeeded in detecting faces based on the specified algorithm. In the experiments, CNN gives better detection accuracy than Haar Cascade. However, for the client application face detection in some specific devices, haar cascade can be used wider than CNN face detection. In mobile, wearable, and Raspberry Pi devices, CNN is hardly used since its GPU requirements. Depends on the This means that the designed application can work and can be used for the next steps for developing the Cloud-based face recognition application.