Introduction In order to identify the faces of the perpetrators from image sequences, image-based face recognition employing probabilistic appearance manifolds was applied. Each criminal’s face is represented by a low-dimensional appearance manifold in the ambient image space, a complex nonlinear appearance manifold expressed as a group of subsets (referred to as posture manifolds), and the cone ivity between these subsets. An affine plane provides a rough estimate of each posture manifold. Exemplars from videos are taken and clustered using a KNN method to create this representation, Each cluster is then represented as a plane calculated using principal component analysis (PCA). The connection between the pose manifolds, which is learned through training video sequences, encodes the probability of transition between images in cach pose manifold, A maximum a posteriori formulation for face is proposed. Problem Identification The current methodologies employed in criminal detection often rely on traditional methods, which may have limitations in accuracy and speed, Identifying potential criminals swiftly and accurately is a critical challenge faced by law enforcement agencies globally. This research addresses the need for a more sophisticated and automated approach to criminal identification, with a specific focus on leveraging the power of CNNs for facial recognition. By doing so, the aim is to overcome the drawbacks of existing systems and provide law enforcement with a reliable tool that can aid in the prompt and accurate identification of individuals with a criminal record. Literature Survey Title: Automated Criminal Identification System using Face Detection and Recognition Authors: Piyush Chhoriya ‘As the world has seen exponential advancement over the last decade, there is an abnormal increase in the crime rate and also the number of criminals are increasing at an alarming rate, this leads toward a great concern about the security issues. various causes of theft, stealing crimes, burglary, kidnapping, human trafficking etc. are left unsolved because the availability of police personnel is limited, many times there is no identification of the person who was involved in criminal activities. To avoid this situation an automated facial recognition system for criminal identification is proposed using Haar feature-based cascade classifier. This paper presents a real-time face recognition using an automated surveillance camera. This system will be able to detect and recognize face automatically in real-time.Title: Criminal Identification by using Real Time Image Processing Authors: Tiwari Aanchaladevi S,Ghotekar Shubhangi $, Ghotekar Ashwini S, Manjare Mayuri A, Pro.Bankar A.A This paper helps in real time face recognition using a automated face surveillance camera, The proposed system consists of 4 steps, it include training of real time images face detection using Haar based classifier comparison of trained real time images with images from the surveillance camera result based on the comparison between them. An most important application of interest is automated surveillance, the objective of automated surveillance to recognize people ‘who around watch list. The ambition of this paper is to compare an image with several images which has been already trained. In this paper, we represent a methodology for face detection strongly in real time environment. Haar cascading is one of the algorithm for face detection. In that we use Haar like classifiers to track faces on OpenCV platform. The correctness of the face recognition is very high. The proposed system can successfully recognize more faces which is useful for quickly searching suspected persons as the computation time is very low. In India, we havea system to recognizing citizen called Aadhaar. If we use this as acitizenship database then we can differentiate between citizen and foreigner as well as we will be able to investigate whether the identified person is criminal or not. Titile: Face Recognition System using OpenCV Authors: Arun Binoy, Acsahmol Edwin, Jayakrishnan K, Pranav S The face recognition system using OpenCV is a system which recognizes know faces and store it with time stamp. Then recognize the unknown faces and store it in another dataset with time stamp. The versatile face recognition system utilizing OpenCV, designed to streamline attendance tracking in online classes, meetings, and residential security applications Leveraging advanced image processing techniques, the system identifies and verifies participants’ faces in real-time, Its adaptable nature allows easy customization of recognized individuals, facilitating efficient management of attendance lists. This technology proves beneficial for remote learning scenarios, virtual meetings, and enhancing security protocols in residential settings. By seamlessly integrating with existing platforms, the system offers a user- friendly and efficient solution to automate attendance management, thereby improving overall convenience and security in various contexts.Objectives Data Collection: © Gather a large dataset of facial images. This dataset should include images of individuals who are known criminals and a diverse set of images for non-criminals. Ensure that the dataset is balanced and representative of the population, Data Preprocessing: © Resize and standardize the facial images to a common size. * Normalize pixel values to a specific range (¢.g.,[0, 1] or [-1, 1). * Augment the dataset to increase its size and improve model generalization (e.g., rotation, flipping, and scaling). Labelling: * Assign labels to the images indicating whether the person is a criminal or non-criminal Model Architecture: * Design a CNN architecture suitable for face identification. Common architectures include VGG, ResNet, or custom-designed networks. © The input layer should correspond to the size of your preprocessed images. The output layer should have a single neuron with a sigmoid activation function for binary classification (criminal or non-criminal).. Training: © Split the dataset into training and validation sets. # Train the CNN using the training set and validate its performance on the validation set. * Utilize binary cross-entropy as the loss function and an optimization algorithm like Adam. Hyperparameter Tuning: + Experiment with hyperparameters such as leaming rate, batch size, and architecture to optimize the model's performance.Methodology This research employs a systematic and ethical methodology to develop a criminal detection system utilizing face identification through Convolutional Neural Networks (CNNs). The overarching strategy encompasses data collection, preprocessing, model development, training, evaluation, and considerations for deployment in the context of advancing technology for public safety. The data collection phase involves a thorough literature review and collaboration with law enforcement agencies to assemble a diverse dataset of facial images. Ethical considerations, including informed consent and anonymization, underscore this process. Subsequently, data preprocessing ensures uniformity by resizing images, normalizing pixel values, and incorporating augmentation techniques to enhance model generalization, Model architecture design draws inspiration from established CNN architectures like VGG and ResNet, with a focus on customization to meet the specific demands of facial identification in criminal detection. The training phase involves splitting the dataset, leveraging GPU resources for efficient model training, and implementing mechanisms such as early stopping to prevent overfitting. Hyperparameter tuning employs systematic experimentation to optimize parameters like learning rate, batch size, and model architecture, Evaluation is conducted on a separate test dataset, with performance metrics such as accuracy, precision, recall, and Fl score informing the model's efficacy in criminal detection. Ethical considerations remain integral throughout the research, addressing biases, engaging with experts, and adhering to privacy guidelines. The deployment and integration phase outlines guidelines for real-world application, collaborating with stakeholders for feedback and refinement. Continual improvement is built into the methodology, ensuring the model remains effective over time by incorporating new data and staying informed about technological advancements. This research methodology aligns with the overarching goal of advancing technology responsibly for the enhancement of public safety through innovative criminal detection systems. Possible outcome * Technical Challenges: Despite advancements in CNNs and facial recognition technology, accurately determining criminality based solely on facial features is highly unlikely due to the complexity and multifactorial nature of criminal behavior. The system may struggle with false positives, misidentifications, and inconsistencies. * Ethical Concerns: There would be significant ethical concerns surrounding such a system. Facial recognition technology is already controversial due to issues like privacy invasion, surveillance, and potential for misuse. Using it for criminal detection exacerbates these concems and raises questions about individual rights, presumption of innocence, and discriminatory practices. + Legal Hurdles: Legal frameworks may pose barriers to the deployment of such a system. Laws regarding privacy, discrimination, and due process would need to be carefully considered. Moreover, legal challenges related to the reliability and validity of using facial features for criminal detection could arise.Features Manage(create/edit/delete) criminals from the database Manage(create/edit/delete) users from the database Search for the criminals in livestream feeds Selection different avatars depending on current needs Leaner ‘ce Eo actin Pa aa en Manage Criminal oes Pns-Lrtan oe SmeeRAINING IMAGES AND SUCCESSFULLY STORING DETAILS IN DATALReferences [1] FaceNet: A unified embedding for face identification and clustering, F. Schroff, D. Kalenichenko, and J. Philbin, 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 815-823.M. Young, The Technical Writer’s Handbook. Mill Valley, CA: University Science, 1989. [2] Security enhancement utilizing motion detection, Signal Processing, Communication, Computing and Networking Technologies (ICSCCN), 2011 Intemational Conference on, T huckalay, 2011, pp. 552-557. T homas, Ashraf, Lal, Mathew, Jayashree, [3] Moghaddam, B,, Pentland, A.P.: Face recognition using view-based and modular eigenspaces. In: Automatic Systems for the Identification and Inspection of Humans, vol. 2277, pp. 12-22 (1994),Sawhney, S., Kacker, K., Jain, S., Singh, S. N., Garg, R. (2019, January). Real-time smart attendance system using face recognition techniques. In 2019 9th international conference on cloud computing, data science engineering (Confluence) (pp. 522-525). IEEE [4] Kodali, R.K., Jain, V., Bose, S., Boppana, L.: 1oT based smart security and home automation system, In: 2016 International Conference on Computing, Communication and Automation (ICCCA), pp. 1286-1289. IEEE (2016) [5] M. Turk and A. Pentland, “Eigenfaces forRecognition,” in Jounal of cognitive neuroscience, vol 3, pp. 71-86, Jan 1991 [6] Roqueiro, Petrushin, Counting people using video cameras, Department of Computer Science, University of Illinois at Chicago, Chicago, IL 60607, USA, 2006. [7] ]Kartik, J.S., Kumar, K.R., Stimadhavan, V.S.: Security system with face recognition, SMS alert and embedded network video monitoring terminal. Int. J. Secur. Priv. Trust Manag. (UISPTM) 2, 15-17 (2013) [8] ] Chin, H.: Face recognition based automated student attendance system. Diss. UTAR (2018) [9] Ramanan, D., Zhu, X.: Face detection, pose estimation, and landmark localization in the wild. In: Proceedings of the 2012 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 2879-2886 (2012)