Smart Attendance Marking System using Facial Recognition

 

Neela Ashish Kumar, Swarnalatha P, Prudhvi Chowdary, Jai Krishna Naidu, Kailasa Sandeep Kumar

Undergraduate Students, School of Computer Science and Engineering, Vellore Institute of Technology, Vellore

 

Abstract:

Nowadays, face recognition has become one of the key research regions in computer technology. Due to this scientists share a keen interest in this area. In this paper, we proposed an application of facial recognition in attendance marking system. The attendance marking system is a difficult process if it is done manually when there is a huge classroom with more number of students which is time-consuming. By using this facial recognition system we can avoid the issues of fake attendance, proxies and save time. In previous facial recognition attendance systems, there were some drawbacks such as intensity of light, head pose etc. The main research of this paper is to overcome those problems and provide the most accurate attendance marking system. This system which is primarily based face detection and recognition algorithms mechanically detect the student and mark the attendance when he enters the school room. The device architecture and algorithms utilized in each stage are defined on this paper. We used algorithms such as Viola-Jones for face detection, PCA for feature selection and SVM for classification. When as compared to conventional attendance marking system the proposed system saves time and effective way to maintain attendance and records of students.

 

 

INTRODUCTION:

In this modern era of automation many scientific innovations have been taken place to save labor and human intervention. In the same way, facial recognition is one of the key areas of development in the field of computer science and it is the most efficient biometric technique for identifying people. We can use this technique in education field as automated face recognition attendance marking system which is way different from traditional attendance marking system as it is time consuming. Even Though it is much easier to install face recognition system in a large setting, the actual implementation may be very hard because it needs to account for all possible appearance variation caused by a change in illumination, facial features, variations in pose, image resolution, sensor noise, viewing distance etc. Many facial recognition attendance systems have been developed each has its own strengths and weakness. The main aim of this research is to overcome those problems. The modern development of machine learning and neural networks and its applications in face recognition helps the process more accurate and efficient. The attendance marking system generally consists of following steps such as Image Acquisition, Databasedevelopment, Pre-processing, Face-detection, Feature extraction and classification and at last, we end with post-processing stage. In the very first stage of attendance recognition system, we detect the faces of students using a voila-jones algorithm and process it through histogram normalization through 100x100size of a pixel, pre-process it finally store it in the database. A vast number of algorithms have been proposed for face detection i.e. AdaBoost algorithm, the Float Boost, Neural Networks, Bayes classifier and the Support Vector Machines (SVM). The total efficiency of the face recognition attendance system algorithm can be increased with the fast face and robust detection algorithm. Here we used we used algorithms such as Viola-Jones for face detection, PCA for feature selection and SVM for classification. The subsequent sections in this paper consist of a literature survey, the detailed implementation of the proposed model and finally results, conclusion, and scope for future development.

 

LITERATURE SURVEY:

In This paper [4] the research is mainly focused on the developing of an automated attendance system with the output having an audio in the lecture or any classroom session so that the faculty can record the student’s attendance. It will be very helpful for the faculties as it saves most of the time as well as the effort especially if there are a huge number of students in a class. If there are a number of students in a class it captures all of the student’s faces and takes the attendance only the faces which are identified. This system shows the use of facial recognition technology and how it will be helpful for the attendance marking. It can also be useful for the exam related issues. But it cannot identify the faces which are similar and have similar facial features

 

Here [5] the author proposed a model for face detection attendance system. Here the system is based on two databases one is student database and another one is the storage database. Firstly the process starts with capturing the images of all the students in the class and images of all the student's related masks will be calculated by the facial points of the images i.e, nose, lips, eyes mainly. Another database in the system is the attendance database. It uses to mark the attendance of the particular student. A camera will be installed at the front of the classroom at some point such that it can cover the whole classroom. Once the image gets captured noise from the images will be removed. Gabor filters will be applied to that images after that so that facial fiducial points will be calculated and also the facial features.

 

In this paper [6] they proposed a model for face detection automated attendance system. This system is mainly focused on the face detection and recognition algorithms. This system works very efficiently. It can easily recognize the students when they enter the classroom. If it matches with the any of the database images he/she will be marked as present for the particular session or else they will be marked as absent. This system mainly consists of the feature extraction, preprocessing, image acquisition, databasedevelopment and the classification stages and then the post-processing stage. These stages are the main stages in the automated attendance marking system.

 

Hemanth Kumar Rathod [7] proposed a conventional method of taking the attendance can be done manually either by the teacher or the lecturer. This process requires a lot of time and effort. It can’t guarantee that there will be no proxies in the class and any other errors. So to overcome these type of issues a biometric feature like facial recognition has been found that has a lot of phases like image acquisition, face detection, feature extraction, face classification etc. The attendance can eventually be marked. In this system, we have used the algorithms like voila jones and the classifiers like SVM so as to get the desired results. By using this system we can take the attendance of the student without any effort from the teacher or the lecturer. But it needs the elements like camera, laptop or the computer and the local network. This method is very easy to use, most reliable and secure.

 

In this paper [8] they proposed a model for attendance management system. This system requires various techniques which are already existed and can use with some modifications. The first step is to design such a system on the facial recognition. The functioning of the algorithms needs to be understood so that the face detection dynamics can be understandable. This system can work most efficiently as we are using the engine's face detection algorithm. It actually divides the image into smaller parts. It will be easy to compare the smaller parts rather than covering the whole face. By the implementation of the algorithm, we were able to track the face of the student. It can be done three ways i.e, head tracking, facial feature tracking, complete tracking. It has high accuracy and can be used multiple times. It will be very useful in the online examinations and also for marking the attendance in the classrooms automatically.

 

In this paper [9] they proposed a model for real-time face detection and recognition. It consists of mainly the camera that will be used to capture the images of the students that will be installed in the classrooms to capture the video frames and for capturing the multiple faces. Those captured faces will be cropped and will be converted to the grayscale so that the number of bits will be reduced that needs to be processed. Then the faces will be compared with the images which are already stored in the database. If it matches with the any of the database images it will be marked as present. It includes several phases like database creation, HOG features, facedetection, SVMclassifier, recognition, attendance. When the system finds the image it captures the image a lot of times continuously. It estimates the attendance of each and every student and finds the best-localized image. With this system, the teacher or the lecturer will find it easy to take attendance without any extra effort and the cost. It needs the elements of thecamera, laptop, and the network connection. It is more secure and reliable and easy to use.

 

PROPOSED MODEL:

The main objective of this face detection automated smart attendance system is to provide a system that is reliable, practical and eliminates time loss and disturbance in traditional attendance systems.

 

The proposed model is based on face detection and recognition algorithms. The system architecture of proposed model is shown below.

 

Fig-1: System Architecture for Face recognition Attendance System

 

For implementing the automated face recognition system we have following algorithm to be considered as shown below

 

A. ALGORITHM:

Pseudo Code for the proposed System

1    Capture the Student Photo.

2    Apply the Viola-Jones algorithm for face detection

3    Extracting the Region of Interest [ROI] in the form of Rectangular Box.

4    Converting it intograyscale, applying the Histogram Equalization and Resizing it to 100x100 pixels

5    Store the pre-processed Image into Database

6    Applying the Principle Component Analysis [PCA] (For feature Extraction) and SVM (for Classification)

7    ost-Processing

Now we have following stages to implement the proposed algorithm for face Detection and recognition

 

IMAGE ACQUISITION:

The camera is placed in the classroom at a distance to capture the images of students. This captured image is given as an input to the system.

 

Fig-2  Extracted and captured faces

 

Face Detection:

A proper and efficient algorithm always enhances the efficiency of the face recognition systems. Various algorithms have been proposed such as Face geometry methods, FeatureInvariant methods, Machine learning based methods etc. Among them, we used Viola and joined detection algorithm which gives high detection rate and also fast. This algorithm is efficient for real-time application for its efficiency. Viola-Jones algorithm is applied to this frame, which detects the faces in the frame. To ensure that the detected object is facing, each detected object is cropped and further processed for eye detection and if eyes are detected they are considered as faces else are rejected. Features of all the faces are extracted using different facial features such as the nose, eyes, the distance between eyes, etc.

 

Fig -3 Face detection for training the data

 

Pre-Processing:

The detected face is extracted and subjected to preprocessing. This pre-processing step involves with histogram equalization of the extracted face image and is resized to 100x100. Histogram Equalization is the most common Histogram Normalization technique. This improves the contrast of the image as it stretches the range of the intensities in an image by making it clearer.

 

Database Development:

This database development phase consists of image capture of every individual and extracting the biometric feature, in our case, it is facing, and later it is enhanced using pre-processing techniques and stored in the database. In our project, we have taken the images of individuals in different angles, different expressions and also in different lighting conditions.

 

Feature selection and Extraction:

Now the feature selection and extraction of the captured image is the important module of the total attendance recognition system. For a reliable, compatible and error-free face recognition system we need good feature extraction attributed and good classification and detection algorithm. The performance of a face recognition mainly depends upon good classification algorithm. Now here we used Viola-jones Algorithm for face detection as it is the most accurate face detection algorithm for real-time application. The main characteristics of the viola-jones algorithm are its very high detection rate and very low false positive rate. Here we take the implementation of HAAR features. The few properties are very common to human faces such as eye region is darker than the upper cheeks and nose bridge is brighter than the eyes. Here we use PCA for feature extraction and SVM for pattern classification.

 

B.  PRINCIPAL COMPONENT ANALYSIS:

It is commonly known as the method of selection of the features as well as the reduction of the dimensions. It will be used for the extraction of principal components of multidimensional data. It was the first algorithm which is used for the representation of the faces economically. In PCA the images will be captured by the camera and by using the eigenfaces the captured images will be shown along with the corresponding projections of the image. Only the meaningful dimensions will be considered instead of all the dimensions of the image. The image will be represented mathematically using PCA as

           

      χ = WY + µ

 

Here the face vector is χ, the eigenface vectors are Y, the feature vector is W and the average face vector is µ.In the recognition of the faces, these will be used as the features for classification. Eigenfaces have introduced early for the use of the analysis of the principal components to solve the problems of the recognition of the faces.

 

C.  SUPPORT VECTOR CLASSIFIER:

Support vector machine classifier(SVM) is a binary classifier. Here the hyper plane will be used like decision function. Regarding the presence of an object such as the human after the training of the images containing the particular object, the decisions will be taken by the SVM classifier. Here the PCA will be used for the extraction of the features and SVM will be used for the classification.SVM has been proposed recently that it is the best and the most effective classifier for the recognition of the patterns. For the recognition of the patterns, SVM finds the closest points which are there in the training set that can be done linearly or non-linearly. The recognition of face has two stages. They are classification and the extraction of the features. The previously mentioned feature extractors joined with classifiers are thought about in different true situations, for example, lighting conditions, Unintentional facial element changes, Expressions.

 

Framework Performance is likewise assessed regarding acknowledgment rate, distance, false positive rate, time which is taken for the preparation. Separation likewise plays as a criterion in this framework display as the picture outlines are caught when an individual goes into the room and face area is resized. So the face area caught at around 4feet and 7feet gives better outcomes. For the Training, data of 150 pictures training time is ascertained.

 

Face Recognition:

After face detection and pre-processing of the captured image feature comparison takes place with respect to the features of the face. So, whenever it detects the face it matches the faces captured in the database if it matches, the attendance is marked. As per the experiment, most of the faces are recognized correctly. Fig 3 shows the recognized face along with the registered number.

 

Fig-4 Face Recognition with Registered number

Post Processing:

In the proposed system, after recognizing the faces of the students, the names are updated into a sheet. The sheet is generated by exporting mechanism present in the database system.

 

D.  GRAPHICAL USER INTERFACE [GUI]:

The GUI is used for face detection and recognition as shown below figures. The first part of the GUI is where we register the student name and Registration number along with the face for training the detected faces.

 

Fig-5 GUI for registration number and Name of Students

 

Now after capturing all the faces as shown in fig 3 the faces should be trained with the algorithm and extracted features for more accuracy of face detection. For training of these faces present in the database as shown in the figure-6 below

 

Fig-6 Captured images in the database after pre-processing

 

We Developed a GUI which traces the images in the database before face recognition after adding to each new face in the database for increasing the accuracy of detection as shown in below figure-7.

  

Fig-7 Training the dataset

 

 

 

RESULTS:

The face detection and recognition system are simple yet efficient than other attendance systems. Here we developed a GUI as shown above to manage the total system. After face registration with the name and Reg no, we extract the detected image and pre-processing with Histogram equations and stored in the database. Now the feature extraction and classification of these images can be done b PCA and SVM as shown above then we train the dataset for accurate results. Finally after detection of the face as shown in fig-8 the attendance is marked and stored in anExcel sheet as shown in the fig-9 below.

 

Fig-8 Showing the student is recognized along with his Reg.no

 

Fig-9 Excel Sheet showing the attendance w.r.t time and date

 

 

PERFORMANCE EVALUATION:

Table 1. PERFORMANCE EVALUATION

 

Fig -10  Performance Analysis of PCA + SVM with other algorithms

 

 

CONCLUSION AND FUTURE WORK:

The smart face recognition system is proven as most efficient attendance system among all other attendance systems due to its consistency, robust and accuracy but also it can have its own limitations. In this system, we used algorithms such as Principal Component Analysis for feature extraction and SVM as classifiers which outperforms other algorithms in real-time applications and proved as best for this attendance systems. The overall system is implemented in python The Future work is to improve the recognition rate with other algorithms as human changes as time such as growing beard, glasses, scarf etc. This system is to developed to recognize 3-4 people at a time and in future we canimprove its recognition rate and achieve greater performance and efficiency.

 

ACKNOWLEDGMENT:

With all respect and gratitude, we would like to thank Prof. Swarnalatha P for guiding us throughout the project and also to all the people who helped directly and Indirectly for completion of our Project successfully.

 

REFERENCES:

1.        M. Turk and A. Pentland, Eigenfaces for recognition, Journal of Cognitive Neuroscience, 3(1), pp. 7186, 1991USA: Abbrev. of Publisher, year, ch.x, sec. x, pp. xxx–xxx.

2.        B. K. Mohamed and C. Raghu - "Fingerprint attendance system for classroom needs", in India Conference (INDICON), 2012 Annual IEEE. IEEE, 2012, pp.433438.

3.        S. Kadry and K. Smaili - "A design and implementation of a wireless iris recognition attendance management system", Information Technology and control, vol. 36, no.3, pp. 323329, 2007.

4.        Poornima, S., Sripriya, N., Vijayalakshmi, B., and Vishnupriya, P. (2017). Attendance monitoring system using facial recognition with audio output and gender classification. International Conference on Computer, Communication, and Signal Processing: Special Focus on IoT, ICCCSP 2017, 0–4. https://doi.org/10.1109/ICCCSP.2017.7944103

5.        Sajid, M., Hussain, R., & Usman, M. (2014). A conceptual model for automated attendance marking system using facial recognition. 2014 9th International Conference on Digital Information Management, ICDIM 2014, 7–10. https://doi.org/10.1109/ICDIM.2014.6991407

6.        Chintalapati, S., & Raghunadh, M. V. (2013). Automated Attendance Management System Based On Face Recognition Algorithms. 2013 IEEE International Conference on Computational Intelligence and Computing Research, 1–5. https://doi.org/10.1109/ICCIC.2013.6724266

7.        Rathod, H., Ware, Y., Sane, S., Raulo, S., and Pakhare, V. (2017). Automated Attendance System using Machine Learning Approach, 0–4.

8.        Malik, R., Kumar, P., Verma, A., & Rawat, S. (2017). Prototype model for an intelligent attendance system based on the facial identification. 2016 International Conference on Information Technology, In CITe 2016 - The Next Generation IT Summit on the Theme - Internet of Things: Connect Your Worlds, 40–43. https://doi.org/10.1109/INCITE.2016.7857586

9.        Balcoh, Naveed Khan, et al. "Algorithm for efficient attendance management: Face recognition based approach." JCSI International Journal of Computer Science Issues 9.4 (2012).

10.      Akshara Jadhav, Akshay Jadhav Tushar Ladhe, Krishna Yeolekar. (2017) Automated attendance system using face recognition.

11.      Priyanka Wagh, Jagruti Chaudhari, Roshani Thakare, Shweta Patil. Attendance System based on Face Recognition using Eigenface and peA Algorithms.

12.      Shehu, Visar, and Agni Dika. "Using real-time computer vision algorithms in automatic attendance management systems." Information Technology Interfaces (ITI), 2010 32nd International Conference on. IEEE, 2010.

 

Received on 03.04.2018       Modified on 21.06.2018 Accepted on 14.07.2018      ©A&V Publications All right reserved Research J. Science and Tech. 2019; 11(2):101-108. DOI: 10.5958/2349-2988.2019.00016.0