International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 03 | Mar 2024
p-ISSN: 2395-0072
www.irjet.net
OBJECT IDENTIFICATION IN DIGITAL FORENSIC IMAGE ANALYSIS V.MOUNICA1, M.SAMUEL FINNY2, V.LEELA PAVAN KALYAN3 ,P.RATNA SAI TEJA4 , K.MANOHAR5 1 Assistant professor MTech(Ph.D.) , Dept of Artificial Intelligence and Data Science and Engineering, Vasireddy
Venkatadri Institute of Technology, Nambur, Andhra Pradesh, India
2-5 UG Students, Dept of Artificial Intelligence and Data Science and Engineering, Vasireddy Venkatadri Institute of
Technology, Nambur, Andhra Pradesh, India ---------------------------------------------------------------------***--------------------------------------------------------------------characteristics, the proposed model endeavors to optimize Abstract - Object identification holds significant
the efficacy of crime investigations while also contributing to crime deterrence via proactive surveillance measures. The document offers an overview of pertinent literature, addresses the challenges inherent in manual data processing, and delineates the sequential methodology and essential algorithms integral to the deployment of the proposed model.
importance in digital forensics for validating media integrity and expediting criminal inquiries. This study introduces an innovative method employing YOLOv8, an advanced object detection model, to automate the identification of diverse items such as firearms, blades, bloodstains, and authentic and forged facial images within digital content. Through the utilization of machine learning techniques, specifically Convolutional Neural Networks (CNNs), the proposed model aims to streamline manual data processing in digital forensics, thus reducing investigation time. Furthermore, this initiative not only hastens the investigative process but also aids in crime prevention by bolstering security in prominent venues through ClosedCircuit Television (CCTV) surveillance. The manuscript deliberates on the challenges associated with manual digital data processing, evaluates existing literature on machine learning for crime detection and object recognition, and delineates a systematic approach for implementing the proposed model. Additionally, it provides an overview of fundamental algorithms, such as CNNs, YOLOv8, supervised learning, elucidating their roles, importance, and potential utilities. This project showcases a promising avenue for intelligent digital data processing in criminal investigations and suggests future avenues for refining security protocols and confronting emerging challenges.
1.1 Literature Review Challenges in manually processing digital data for crime investigation are manifold and can significantly impede the investigative process. Firstly, the manual handling of digital data is inherently time-consuming and labor-intensive. Investigators often need to sift through vast amounts of data, such as images, videos, and documents, which can consume considerable time and resources. Moreover, manual processing suffers from limited scalability and efficiency. As the volume of digital data continues to grow exponentially, manual methods struggle to keep pace, leading to bottlenecks in investigations. This limitation hampers the timely analysis of evidence, potentially delaying crucial breakthroughs in criminal cases. Additionally, manual processing is susceptible to human error and bias. Investigators may inadvertently overlook important details or misinterpret evidence, leading to flawed conclusions. Furthermore, human biases can influence decision-making during the analysis process, potentially compromising the integrity and objectivity of the investigation.
Key Words: Digital forensics, CNN, Fake faces, Real faces, YOLOv8.
1.INTRODUCTION Digital forensics plays a crucial role in contemporary law enforcement, serving to authenticate media, probe criminal activities, and safeguard digital evidence. However, the manual handling of extensive digital datasets presents formidable obstacles, often leading to prolonged and laborious investigations. With the evolution of machine learning methodologies, there exists an opportunity to streamline this procedure by automating object recognition in digital content. This manuscript introduces a project dedicated to harnessing YOLOv8, an advanced object detection model, alongside machine learning algorithms to expedite investigative processes in digital forensics. Through the precise identification of items such as firearms, blades, bloodstains, and facial
© 2024, IRJET
|
Impact Factor value: 8.226
Addressing these challenges requires the adoption of automated tools and techniques, such as machine learning algorithms and object detection models. These technologies offer the potential to streamline data processing, enhance scalability and efficiency, and mitigate the risk of human error and bias. By automating object identification and analysis, investigators can expedite the investigative process, improve the accuracy of results, and ultimately facilitate more effective crime investigation and resolution.
|
ISO 9001:2008 Certified Journal
|
Page 1054