Enhancing University Security: A Machine Learning and IoT Driven Face Recognition System for Surveillance and Attendance

doi:10.21203/rs.3.rs-3130718/v1

Download PDF

Research Article

Enhancing University Security: A Machine Learning and IoT Driven Face Recognition System for Surveillance and Attendance

https://doi.org/10.21203/rs.3.rs-3130718/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Convolution Neural Networks (CNN) doesn't need any techniques to implement feature selection, extraction and learns the data based on the optimizer, layers, increasing or decreasing learning rate. For design of face recognition model we can directly use detection model Convolution Neural Networks (provided by GPU) pre-trained or HOG for faster results without any training. The images in the dataset are cropped based on the region of interest where the image gets saved only if it detects eye and the embeddings. The dataset is then encoded into NumPy arrays and stored in a pickle file for prediction. The encodings are used to detect faces over image, a live video and over a video file. To train a network from scratch we need to train a lot of images so using a pre-trained model with fine tuned weights reduces training time and images are convoluted and trained as per the architecture. To run this prototype accurately we can use CNN as a detection model to recognize images as it are accurate in prediction. Design of the Face recognition system can also be done by model training. This technique helps us to increase data if required depending on performance or hyper parameter tuning to increase the efficiency of model accuracy. We used Support Vector Machine (SVM) algorithm where we have sent cropped raw images long with wavelet transformation to train the classifier. So that the model will learn the global features in the image and can accurately predict the outcome with the accuracy of 99.63%.

Convolution Neural Networks (CNN)

Computer vision

Machine Learning

Face Recognition

Internet of Things

Support Vector Machine (SVM)

This implementation has three phases where the first part is to maintain a dataset through which an image can be retrieved with a recognized face [4]. The image should be encoded by encoding() function in python by NumPy array and saving it in a pickle file which converts encodings to make sure whether the person is the required target to be recognized. Since each image cannot be encoded every time argparsing is quite a good technique where a query can execute and give the required output which the user requires. Many images can be encoded and stored in the array in a. pickle extension.

The next phase is to recognize an image in the dataset. To do the task Computer Vision is the main concept to move further in the implementation. Secondly, an image is converted into a frame in which the pickle module provides the interface to recognize the image. The main idea is to read the image converts the image into matrix distribution in a 3D array and that is compared with the dataset and would give the name of the image surrounded by the box [5]. Functions like imread() are used to read the image, imshow() to show the image from CV2 and finally, the encodings are matched with the names with the required index. A dictionary is set to count the total number of times each face was matched with the given encodings and thus would provide the required output image.

This is the challenging phase where the image should be recognized over a video stream. The same encodings should be matched with the following video stream where the face in the image is recognized. Since the video streaming requires high-quality Webcam/D3D CC Camera, therefore code is changed to produce frames and recognition of image becomes easy. Through this module called imutils the video stream breaks the video into the frame for every second and recognizes the name of the person from different angles.

Authentication checking in a university surveillance and attendance system using face recognition based on machine learning and the Internet of Things (IoT) involves a combination of technologies to ensure accurate identification and verification of individuals. Face Detection: The system uses computer vision techniques to detect and locate human faces in real-time. This step is essential for subsequent face recognition. Face Recognition holds machine learning algorithms, such as convolution neural networks (CNNs), are trained on a dataset of facial images to recognize and extract unique features from each face. These algorithms create face embeddings or representations that can uniquely identify individuals [6]. Enrolment initially deals students and staff members would need to enrol in the system. During enrolment, individuals' facial images are captured and associated with their personal information, such as names, IDs, and other relevant details.

This information is stored securely in a database. Real-time Recognition deals once the system is deployed, cameras placed strategically within the university premises capture live video feeds. The system analyzes the video frames, detects faces, and matches them against the enrolled faces in the database. This step verifies the identity of the person in real-time. Authentication and Attendance Tracking describes when a recognized face is detected, the system checks the corresponding database entry to authenticate the person's identity. If the authentication is successful, the system marks the individual as present and records the timestamp for attendance tracking purposes [7].

Internet of Things (IoT) Integration describes to make the system more efficient and scalable, IoT devices can be incorporated. For example, IoT sensors can detect individuals entering or leaving specific areas (e.g., classrooms or labs) and trigger the face recognition process automatically. This integration ensures real-time and accurate attendance data collection. Reporting and Monitoring shows the system can generate attendance reports and provide real-time monitoring of attendance data. Administrators, teachers, or authorized personnel can access this information through a secure web interface or mobile application. They can monitor attendance, analyze patterns, and generate reports as required. It's important to note that privacy concerns and ethical considerations must be addressed while implementing such a system. Clear guidelines on data collection, storage, and usage should be established to protect individuals' privacy rights and ensure compliance with relevant regulations. Additionally, regular maintenance, system updates, and staff training are necessary to maintain the system's accuracy and reliability [8].

2.1. Algorithm for CNN model: The general description of algorithm follows

Step.1. Split the dataset into training and validation sets.

Step.2. Design a CNN architecture suitable for face recognition, which typically includes multiple convolution layers, pooling layers, and fully connected layers.

Step.3. Train the CNN model using the training set and optimize it using techniques like back propagation and gradient descent.

Step.4. Fine-tune the hyper parameters of the model (e.g., learning rate, batch size) to achieve better performance.

Step.5. Evaluate the model using the validation set and iterate on the architecture or hyper parameters as necessary.

The above algorithm provides a high-level overview and may require further customization based on the specific requirements, available resources, and technological constraints of the university's surveillance and attendance system.

2.2. Architecture of CNN:

Designing an easy-to-use and low-cost can be achieved through a combination of hardware and software components. Hardware Components includes Cameras; Use affordable IP cameras or webcam devices placed strategically in classrooms, entrances, and other key areas to capture images or video streams. IoT Devices deploy IoT devices such as Raspberry Pi or Arduino boards to connect cameras and other sensors to the network [9]. Network Infrastructure will set up a reliable network infrastructure to ensure seamless connectivity between devices. Figure.1. shows Methodology for Face Recognition and figure.2. Shows an architecture of University Surveillance and Attendance System.

University surveillance and security are important aspects of maintaining a safe and secure environment on campus. Educational institutions often implement various measures to ensure the well-being of students, staff, and visitors. Security Personnel shows many universities employ security personnel, such as campus police or security officers, who are responsible for patrolling the campus, monitoring activities, and responding to incidents [10]. These individuals are trained to handle security-related issues and can provide a visible presence to deter potential threats. Access Control describes universities often implement access control systems to regulate entry to buildings and sensitive areas.

3.1 University Surveillance and Security:

Video Surveillance indicates closed-circuit television (CCTV) cameras are commonly used to monitor key areas of the campus, such as entrances, parking lots, and common spaces. Video surveillance can assist in deterring criminal activities, identifying suspects, and providing evidence in case of incidents. Emergency Communication Systems shows universities often have emergency communication systems in place to quickly disseminate information during critical situations. This may include text alerts, email notifications, public address systems, and mobile applications. Effective communication systems help to alert the campus community about potential threats, emergency situations, or evacuation procedures. Campus Lighting shows adequate lighting is essential for maintaining a safe environment on campus. Well-lit pathways, parking lots, and outdoor spaces can deter criminal activities and enhance personal safety, particularly during night time. Regular maintenance and assessment of lighting infrastructure are crucial to ensuring its effectiveness. Crime Prevention Programs shows universities often conduct crime prevention programs to educate students and staff about personal safety, crime prevention strategies, and reporting procedures. These programs may include self-defence classes, safety seminars, and workshops on topics like sexual assault prevention, alcohol and drug abuse awareness, and cyber safety [11].

This collaboration can involve sharing information, coordinating responses to incidents, and conducting joint training exercises. Student Involvement describes encouraging students to actively participate in maintaining a secure campus environment can be beneficial. This can include initiatives like campus watch programs, where students are trained to report suspicious activities and act as additional eyes and ears for security personnel [12]. Threat Assessment Teams leads universities may establish threat assessment teams comprising representatives from various departments, including security, counselling services, administration, and law enforcement. These teams assess and respond to reports of concerning behaviour, helping identify and intervene in potential threats to campus safety. Privacy Considerations describes while ensuring security is important, universities must balance it with privacy considerations. It's crucial to handle and store surveillance data responsibly, limit access to authorized personnel, and comply with applicable privacy laws and regulations. It's worth noting that specific security measures and approaches can vary between universities depending on factors such as location, campus size, and perceived risks. Universities should regularly assess their security systems, conduct drills and simulations, and adapt their approaches based on the evolving nature of threats and best practices in the field.

3.2. Automatic Student’s Attendance System

An automatic student attendance system is a technology-based solution that automates the process of tracking and recording student attendance in educational institutions. It replaces traditional manual methods such as taking roll call or using sign-in sheets. Identification of the system typically uses some form of identification method to recognize individual students. This can include options like student ID cards with embedded RFID tags, biometric recognition (such as fingerprints or facial recognition), or even mobile apps with unique login credentials [13]. Where data capture: describes once the student's identification is verified, the system captures the attendance data. For example, if RFID tags are used, a scanner or reader records the student's presence by detecting the tag when they pass by. In the case of biometric recognition, the system captures the necessary biometric data for attendance tracking. Moreover, data processing deals the attendance system processes the captured data and converts it into a usable format.

It associates the attendance data with specific students and timestamps. Storage and Management will store in a centralized database or cloud system, which allows for easy access, analysis, and management of attendance records. This data can be used for generating reports, calculating attendance percentages, and identifying patterns or trends [14]. Integration and Reporting: The automatic attendance system can integrate with other software systems used by educational institutions and this integration enables seamless data exchange and facilitates the generation of attendance reports for teachers, administrators, and parents. Benefits of an automatic student attendance system holds efficiency and it eliminates the need for manual attendance taking, saving time for both teachers and students [15]. The process is streamlined and automated, reducing administrative burdens. Accuracy may takes the use of technology minimizes the chances of human error or manipulation. Attendance records are more reliable and precise [16]. Real-time Monitoring system will provides real-time attendance information, allowing teachers and administrators to monitor student attendance promptly [17]. This enables timely intervention for students who may be consistently absent or tardy. The following Figure.3. describes (a). Image Dataset-1, (b). Image Dataset-2, (c). Image Dataset-3, (d). All the faces in the image are trained to show 128-d embeddings trained in CNN and HOG method [18]. This information can aid in identifying and addressing attendance-related issues [19]. Improved Communication describes the system can be integrated with communication tools to send automated notifications to parents or guardians regarding their child's attendance, ensuring prompt communication and collaboration [20].

It is important to note that the implementation of an automatic student attendance system should consider privacy and data protection regulations to safeguard the personal information of students. Institutions should adhere to relevant guidelines and obtain appropriate consent when implementing biometric recognition or using personal data for attendance tracking purposes. Where Figure.4 shows Faces are recognised through CNN method (a) CCTV footage, (b). Through Webcam short distance, (c). Through Webcam long distance (d) Recognition through Video. Figure.5. describes (a) Original Image, (b) Pre-processed Image, (c). Confusion Matrix for face Recognition.

The implementation of the University Surveillance and Attendance System also enhances campus security. With the ability to monitor and track individuals within the premises, the system can swiftly identify unauthorized individuals or potential threats. This proactive approach helps create a safer environment for students, faculty, and staff. Additionally, the system offers convenience and time-saving benefits for all stakeholders. Students can easily mark their attendance by simply presenting their faces, eliminating the need for cumbersome manual processes. Faculty members can also benefit from streamlined attendance management, as the system automatically generates accurate attendance reports, saving valuable time and effort. In summary, the system presents a compelling solution for improving attendance tracking and enhancing security within academic institutions. By leveraging the power of advanced technologies, universities can streamline administrative processes, improve accuracy, and create a safer environment for all members of the campus community. Whereas the Support Vector Machine (SVM) plays an important role in face recognition and it has the accuracy of 99.63% when compared with all previous techniques and methods.

Funding statement: The authors have received no specific funding for this work.

Ethical approval: This article does not contain any studies with human participants performed by any of the authors.

Conflict of interest: The author has no relevant financial or non-financial interests to disclose.

Data Availability statement: All relevant data are within the paper and its Supporting Information files.

Bhatt, R., & Patel, V. (2019). Smart attendance system using IoT and facial recognition. In 2019 2nd International Conference on Intelligent Computing, Instrumentation and Control Technologies (ICICICT) (pp. 826-831). IEEE.
Parate, V., & Dubey, V. (2020). IoT based student attendance monitoring system using face recognition. In 2020 2nd International Conference on Advances in Electrical, Computing, Communications and Sustainable Technologies (ICAECT) (pp. 37-41). IEEE.
Chen, C., Chen, C., & Lai, C. (2018). The implementation of face recognition based on machine learning in the classroom attendance system. In 2018 International Symposium on Computer, Consumer and Control (IS3C) (pp. 231-234). IEEE.
Sharma, A., & Thirumalai, R. (2018). A review on smart attendance system using IoT and machine learning. In 2018 3rd IEEE International Conference on Recent Trends in Electronics, Information & Communication Technology (RTEICT) (pp. 2166-2170). IEEE. [5].
Sreekumar, M. V., & Nair, R. R. (2020). Face recognition based attendance monitoring system using IoT. In 2020 International Conference on Smart Electronics and Communication (ICOSEC) (pp. 13-16). IEEE.
Chavan, S., & Pawar, S. (2019). Attendance monitoring system using face recognition based on IoT. In 2019 International Conference on Advanced Computation and Telecommunication (ICACAT) (pp. 110-114). IEEE.
Nguyen, N. H., Hoang, D. T., Nguyen, H. D., Nguyen, D. C., & Nguyen, N. H. (2020). An intelligent student attendance system using IoT and facial recognition in a university environment. Journal of Ambient Intelligence and Humanized Computing, 11(8), 3541-3551.
Taha, K., Fathy, M., Roushdy, M., & Tolba, A. (2020). IoT-based smart classroom for monitoring students' attendance using face recognition. In 2020 2nd International Conference on Computer Applications & Information Security (ICCAIS) (pp. 1-5). IEEE.
Keshari, M., Jain, S., & Sahu, N. (2021). An efficient face recognition-based attendance system for universities using machine learning and IoT. In 2021 8th International Conference on Signal Processing and Integrated Networks (SPIN) (pp. 238-243). IEEE.
Yadav, A., Kumar, A., & Jaiswal, D. K. (2021). An automated attendance system using IoT and facial recognition for university applications. In 2021 International Conference on Emerging Trends in Information Technology and Engineering (ICETITE) (pp. 1-6). IEEE.
Kaur, H., Dhingra, P., & Gupta, S. (2022). Facial recognition-based attendance system using deep learning and IoT for universities. In 2022 4th International Conference on Inventive Systems and Control (ICISC) (pp. 1791-1794). IEEE.
[12]. Gudikandhula Narasimha Rao, et al, “Fire detection in Kambalakonda Reserved Forest, Visakhapatnam, Andhra Pradesh, India: An Internet of Things Approach”, Journal of Materials Today: Proceedings, Elsevier, Volume 5, Issue 1, Pages: 1162–1168, 2018, ISSN: 2214-7853
Rao, G.N., Rao, P.J., Duvvuru, R.: A drone remote sensing for virtual reality simulation system for forest fires: semantic neural network approach. In: IOP Conference Series: Materials Science and Engineering. vol. 149. no. 1. IOP Publishing, 2016.
Ahmed, I., Sharif, A., & Ilyas, M. (2022). A comprehensive survey on face recognition-based attendance systems using deep learning and IoT. In 2022 4th International Conference on Communication, Computing and Digital systems (C-CODE) (pp. 119-123). IEEE.
Rao, Gudikandhula Narasimha, et al. "Neural fuzzy system design in forest fire detection." Microsystem Technologies (2023): 1-13.
Rao, G. N., Rao, P. J., Duvvuru, R., Beulah, K., Rathnala, P., Lydia, E. L., ... & Motru, V. R. (2023). Geospatial Study on Forest Fire Disasters--A GIS Approach. Ecological Engineering & Environmental Technology (EEET), 24(1).
F. V. Massoli, G. Amato, F. Falchi, C. Gennaro and C. Vairo, “CNN-based System for Low Resolution Face Recognition” Conference: 27th Italian Symposium on Advanced Database Systems, AIMIR Research Activities 2019. (2019)
K. He, X. Zhang, S. Ren and J. Sun, Deep Residual Learning for Image Recognition, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, (2016), 770-778.
Joseph, Dona. (2020). Automatic Attendance System using Face Recognition. International Journal for Research in Applied Science and Engineering Technology. 8. 769-773. 10.22214/ijraset.2020.30309.
A. Shetty, Bhoomika, Deeksha, J. Rebeiro and Ramyashree, "Facial Recognition using Haar Cascade and LBP Classifiers", 2021.

Download PDF

Version 1

posted

You are reading this latest preprint version

Enhancing University Security: A Machine Learning and IoT Driven Face Recognition System for Surveillance and Attendance

Status:

Version 1

Abstract

Figures

1. Introduction

2. Materials and Methods

2.1. Algorithm for CNN model: The general description of algorithm follows

2.2. Architecture of CNN:

3. Results

3.1 University Surveillance and Security:

3.2. Automatic Student’s Attendance System

4. Conclusions

Declarations

References

Status:

Version 1