Impersonation detection using face verification & recognition
Impersonation detection using face verification & recognition

The goal of the project was to automate the process to verify and identify persons by matching digital images, photos, video feed with an existing database and need to recognize the fraud involved in the complete process using deep learning neural networks and few pre-trained models which were trained by million faces to extract facial features. This method identifies impersonation cases which happen commonly during physical ability tests for police. Also it detects if there is any discrepancy in the candidate registered to a particular exam and the candidate attending the exam. For a particular government exam, registration photo, photo at preliminary examination and photo at main examination is given as a dataset. A CCTV live streaming during a physical ability test is provided. Based on the designation of the government exam, a certain number of laps have to be completed by the candidate in a given time. We have to detect if the candidate registered and candidate appeared for the physical ability test are the same.