A Deep Learning Based Fast Face Detection and Recognition Algorithm for Forensic Analysis

Abstract

Face detection and recognition applications have become an important research issue with the advent of artificial intelligence and deep learning studies, which have drawn much attention to researchers in recent years. The knowledge of retrieving contents from image, video, and audio files is seen as one of the vital influences in the field of digital forensics. In this study, we proposed a software that is supported with deep learning model which can analyze video and image files extracted from forensic evidence for either face detection or object recognition in a folder given as input and file a report of the images and videos that have been realized via this proposed application. The proposed deep learning model trained with YOLOv5 object detection algorithms can easily detect faces that are completely or partially visible under different light levels or in the image. This study shows that deep learning supported solutions can be easily preferred for real-time applications and time-consuming implementations to reduce fatigue and error-prone incurred during forensic investigations.