Biometric systems have been using widely. These systems are being used for user authentication and verification and have improved substantially. This is playing a key role in personal, national and global security. In today's world, face recognition systems have become increasingly important due to their various applications like surveillance, forensic investigations, access control, etc. However, one of the major concerns with face recognition systems is their vulnerability to face spoofing attacks. Face spoofing detection come in handy with face recognition system. Its purpose is to detect a spoofing attack by identifying real and spoof face. This study is concerned with face spoofing detection for both twin as well as non-twin. To accomplish this study the first task is acquiring the dataset. Due to resource limitation the model undergoes training, validation and testing using a subset of CelebA dataset, which contains 200,000 images of celebrities from various parts of the world. The next task is pre-processing which includes histogram equalization, face detection, and image resize. For those techniques, different algorithms are nominated and better suit for the dataset is selected. For the purpose of feature extraction, GAN, CNN, and LBP are used as feature extractors. GANs are designed to generate new data; however it can also be used for feature extraction and classification task. The utilization of GAN in twins' feature extraction was due to its unsupervised characteristics. This is crucial for twins face spoofing detection, considering that the distinctions between twin faces is challenging and CNN to automatically learn features that are invariant to illumination, pose, and facial expression & LBP to captures the local texture information of a face for individuals’ feature extraction. Features from these feature extractors are combined and fed to GAN model for classification. The last task is classifying images as real or spoof. For classification purpose, we used modified GAN. The developed model registered accuracy of 79%, 77%, and 75% when tested with features extracted by CNN, GAN, and LBP respectively. However, the model achieved 95% accuracy when tested with combined features of these feature extractors. Keywords: CNN, Face spoofing, GAN, LBP, Spoof detection