Face Recognition with Liveliness Detection Lo...

Face Recognition with Liveliness Detection Login on Deep Reinforcement Learning for Face Anti-Spoofing in Attendance System

Author Name : Pushpa Viyyapu

ABSTRACT

Machine learning has recently drawn a lot of interest from clients in a variety of fields, including medicine, disease prediction, object recognition, biometric authentication, image processing, rating prediction, review analysis, and so forth. Though many users are interested in machine learning, it still has limitations in some areas, such as facial recognition, classification, and detection methods. This inspired us to use Deep Reinforcement Learning to enhance the functionality of those applications (i.e., which is a subfield of ML and reinforcement and deep learning). By extracting the features from the facial dataset provided by the client for training the model, the basic deep learning model is quite effective for face recognition, but it may not verify the exterior features like whether the face is real or fake. As a result, it became very difficult for some businesses to verify users using face recognition, so we developed deep reinforcement learning (DRL), which uses both reinforcement and deep learning to gauge how active a user is before authenticating them. The application can be trained on a variety of examples, but for testing purposes, we're going to use facial recognition and user liveliness detection to check the application on an attendance management system and prevent spoofing. This DRL model is primarily employed to address issues with ageing, low resolution, and position fluctuations. By using one algorithm as the base algorithm, the suggested model aims to merge many algorithms to develop the CNN model and other algorithms, and to test user faking. Here, we aggregate the face pixels using Local Binary Patterns Histograms (LBPH) and determine each pixel's liveliness by comparing it to its neighbors. This is a component of computer vision (CV) for recognizing the human face as we load in front of the camera. Along with this technology, we employ active anti-spoofing face recognition liveliness detections, which locate spoof users by analyzing video sequences for features like blinking eyes or pupil tracking.

Index Terms— Convolutional Neural Networks (CNN), Local Binary Patterns Histograms (LBPH), Face Dataset, Computer Vision (CV), Deep Reinforcement Learning (DRL).