Developing a Robust Face Recognition Algorithm with Anti Spoofing Using InceptionV3 and YOLOv8

With the development of technology, also comes the rise of feature-based recognition systems. One such system is the face recognition system. Despite the rapid improvements to the system, it has a risk in that the system can become susceptible to spoofing. Our research aims to address this issue by combining two models, one focusing on face recognition, while the other is focused on detecting any spoofing. Utilizing Inception V3strong feature extraction performance and classification accuracy and YOLOv8, known for its real-time object detection capabilities, we desire to develop a combined model capable of applying accurate face recognition and capable of dealing with spoofing attacks. The algorithm works by first identifying the captured input as real or fake using YOLOv8, once the input is confirmed as real, the process continues with a facial recognition with Inception V3.Results of testing showed that the algorithm performs accurately in both tasks. However, the resulted integration caused a low framerate to be captured due to high computational power requirement. Future works aims to find methods to enhance the efficiency of the model, either by optimization or utilizing a hardware with higher computational power, to hopefully create a robust system that can be used for example a face recognition attendance system.