Author : Yuvraj B. Deshmukh and Dr. S. K. Korde

Forensic video/Image analysis is used in a many visual digital evidence analysis scenarios. Video/Image forensics is important & necessary to show that images and videos which are used as potential evidence in court of law are verifiably true. The popularity of smart digital/mobile devices and due to increasing number the low cost of surveillance systems, several forms of visual data are widely being used in digital forensic investigation. Digital videos are many times used as key evidence sources in evidence identification, analysis, presentation, and report. The motto here is to develop advanced forensic video analysis methods & techniques in order to perform effective digital forensic investigation. Further the aim is to develop a forensic video/image analysis framework that employs an efficient video/image processing techniques using deep learning, such as object detection framework YOLO V3, Head pose estimation technique in wild and face detection, color, face and skin detection methodology, also enhancing algorithm for the low quality of video footage analysis which consists of adaptive video enhancement algorithm based on Contrast Adaptive Histogram Equalization (CLAHE) technique, Contrast Exposure Fusion Algorithm, Dynamic Histogram Equalization (DHE) which are introduced to improve the quality of visual data for the use of digital forensic identification & investigation. The technique of Video/Image tampering detection using state-of-art techniques and textual enhancement are also incorporated in order to assist in truthful analysis of visual data evidence. The framework would deploy recent techniques and algorithms which will assist examiner to perform visual analysis of evidence data.

Keywords: Video/Image Analysis, deep learning, object detection, Enhancement Algorithm, head pose estimation, tampering detection.