Face Recognition Using KD-tree Algorithm for Fraud Detection

Fraud or Malpractice occurs when someone who is not supposed to be in an exam room passes off as someone else. This can lead to students being marked incorrectly, students being given extra points, or failing grades because they were given answers or answers given to them by someone other than their teacher or instructor. The only way we will be able to prevent this sort of thing from happening is through machine learning and AI technology that can identify people based on their appearance and behaviour patterns. Face recognition has been the subject of much research and development. Lately, biometric systems have been improved by incorporating features from object recognition in addition to human behaviour analysis. However, some methods are not effective for large-scale databases. The KD-tree algorithm-based acceleration technique for a large-scale face recognition system is proposed in this paper. Face recognition takes a long time because it needs to process numerous photographs one at a time when working with a huge database. We use the KD-tree algorithm to address this issue. K means algorithm combined with the KD-tree method is another potential solution to the mentioned issue. We use the k-means clustering algorithm to group facial features to be more precise. In particular, the data in each cluster are kept as KD- trees, and closest neighbour searches based on KD trees are used to match faces to features. Studies using self-collected databases demonstrate that our suggested method outperforms other various face recognition algorithms.