On the Detection of Manipulated Identification Documents

As concerts and events have increased in scale, event organizers are forced to rely on ticket management systems to automate the admission to their events. As more event spaces can accommodate more patrons it has become more important to ensure that event spaces are not oversold. Many instances of duplicated and fake ticket admissions have led to catastrophe, violating regulations and creating unsafe situations for patrons. While online ticket sales increase customer convenience, they also increase accessibility for threat actors, which can potentially cause loss for the event organizers. One solution is to require a form of identification to gain admission to an event. However, currently, there exists a significant gap in reliably detecting manipulated identification documents, especially with the rapid improvements in artificial intelligence (AI). These advanced AI models can produce forged results that are so realistic that the artifacts and inconsistencies left behind are almost invisible to the average person. In this paper, we present a method for detecting documents that have been manipulated using face-swapping techniques by employing the power of Convolutional Neural Networks (CNN). This would aid in reducing the losses suffered by event organizers and prevent them from admitting fake or duplicated tickets while improving the convenience to legitimate customers.