Crime Scene Prediction for Unmanned Aerial Vehicles Investigation via Machine Learning

Abstract

Unmanned Aerial Vehicles have increased their applicability in evidence gathering for collecting digital evidence. Authenticating this evidence depends on the crime scenario and the type of drones used. With drones, the evidence type could be the GPS location, the communication signal, the flight log, and the multimedia evidence. These are forms of evidence that forensic experts use to finalize crime cases. Analyzing this evidence requires expensive forensic toolsets. This paper will investigate the level of accuracy and performance verification of machine learning algorithms such as Convolutional Neural Networks (CNN), Support Vector Machines (SVM), and Long Short-Term Memory (LSTM) can reach in identifying vehicle information and the details of the people inside the vehicle at the crime scene and starting with the level of identifying the minute details for the investigation. The performance of the machine learning algorithms will analyze the evidence for deepfakes and possible manipulations in the evidence gathered.