International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 04 | Apr 2024
p-ISSN: 2395-0072
www.irjet.net
ADVANCED FORENSIC FACE SKETCHING AND RECOGNITION Prof. L. Rasikannan1, S. Gnanaprakash2, D. Naveenkumar3, K. Vishnuprakash4 1Associate Professor, Department of CSE, Government College of Engineering, Srirangam, Tamilnadu, India 2,3,4UG student, Department of CSE, Government College of Engineering, Srirangam, Tamilnadu, India
---------------------------------------------------------------------***--------------------------------------------------------------------innovative feature ensures that the created sketches closely Abstract - This paper addresses the challenges faced by resemble hand-drawn ones, facilitating easier adoption by law enforcement departments.
traditional hand-drawn face sketches in forensic art, particularly their time-consuming nature and limited compatibility with modern recognition technologies. We propose a novel application designed to streamline the process of creating composite face sketches without the need for forensic artists. Through intuitive drag-and-drop functionality, users can effortlessly generate sketches of suspects. Moreover, our application incorporates advanced deep learning algorithms and cloud-based infrastructure to facilitate rapid and accurate matching of these sketches with police records. This approach significantly improves the efficiency of criminal identification processes, marking a substantial advancement in forensic investigation techniques.
This paper introduces an application designed to aid law enforcement in suspect identification by integrating advanced deep learning algorithms and cloud infrastructure. Notably, our platform enables the uploading of previous hand-drawn sketches, which are then processed using efficient deep learning techniques. This approach enhances the accuracy and efficiency of suspect recognition, offering law enforcement teams a powerful tool for criminal investigation. Our platform utilizes machine learning algorithms to accelerate the creation of facial sketches by suggesting relevant facial features based on user selection. By learning from both the sketches and the database, the algorithm provides users with a curated list of compatible features, reducing the time required to complete a sketch. This approach significantly enhances the efficiency of the platform, empowering users to generate accurate composite sketches more quickly and effectively.
Keywords: Forensic Face Sketch, Face Sketch Creation, Face Recognition, Criminal Identification, Deep Learning, Machine Locking, Two Step Verification.
1. INTRODUCTION This study focuses on enhancing the process of identifying criminals through face sketches based on eyewitness descriptions. While hand-drawn sketches have been the traditional method, they're often slow and ineffective, especially when matching against existing or real-time databases. To address this, we propose leveraging digital tools to streamline the process. By using technology to create and match sketches, we can significantly improve efficiency and accuracy in identifying criminals, ultimately expediting the path to justice for victims.
1.1 RELATED WORK This study examines advancements in face sketch construction and recognition, focusing on the work of Dr. Charlie Frowd, Yasmeen Bashir, Kamran Nawaz, and Anna Petkovic. Initially, they developed a standalone application for constructing and identifying facial composites, which proved to be time-consuming and confusing. Subsequently, they adopted a new approach where the victim was presented with options of faces resembling the suspect and asked to select similar ones. The system then combined these selections to predict the criminal's facial composite automatically. Promising results were obtained, with 10 out of 12 composite faces correctly named. The study found a success rate of 21.3% when witnesses were assisted by department personnel and 17.1% when witnesses attempted construction themselves.
This study examines past attempts to automate suspect identification through modifications of hand-drawn face sketches, highlighting their limitations in providing accurate results. Despite the introduction of applications for creating composite face sketches, these tools faced challenges such as a restricted range of facial features and a cartoon-like appearance, hampering their effectiveness and efficiency. This paper explores the need for improved techniques to address these shortcomings and enhance the precision of suspect identification from police databases.
Xiaoou Tang and Xiaogang Wang proposed a recognition method utilizing a Multiscale Markov Random Field Model to synthesize photo-sketches. The project aimed to transform sketches into photos and vice versa, enabling database searches for relevant matches. The model divided face sketches into patches and synthesized available photos into sketches, subsequently training the model to minimize
This paper explores the development of an application aimed at improving the creation of face sketches for suspect identification. Unlike previous applications that offered a limited set of facial features for selection, our approach allows users to upload hand-drawn features, which are then converted into components within the application. This
© 2024, IRJET
|
Impact Factor value: 8.226
|
ISO 9001:2008 Certified Journal
|
Page 2068