International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 08 Issue: 12 | Dec 2021 www.irjet.net p-ISSN: 2395-0072 © 2021, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 432 A Comprehensive Review for Smart Attendance Monitoring System Using Machine Learning and Deep Learning Sushma Vispute 1 , Dr. K. Rajeswari 2 , Pratik Adhav 3 , Avadhoot Autade 4 , Abhimanyu BabarPatil 5 , Aditya Dhumal 6 1-6 Dept. of Computer Engineering, Pimpri Chinchwad College of Engineering, Maharashtra, India ---------------------------------------------------------------------***---------------------------------------------------------------------- Abstract - In the research of attendance monitoring systems, it is observed that on average 10-15 minutes of the lecture are wasted on taking attendance using conventional methods. These drawbacks can be solved using a facial recognition-based attendance tracking system. Face recognition can be done using different machine learning, deep learning algorithms. This paper compares various face recognition-based models which use machine learning, deep learning, OpenCV, Internet of Things (IoT) based approaches. Most of the authors used the Haar Cascade algorithm for face detection. Among all the machine learning and deep learning algorithms, Convolutional Neural Networks (CNNs) were found out to be the most accurate and reliable. According to many authors, the accuracy of CNN is found in between 95-98%. Key Words: Haar Cascade, Convolutional Neural Networks (CNNs), Machine Learning, image Processing, Deep Learning, Face Recognition, Attendance monitoring, Internet of Things (IoT) 1.INTRODUCTION In every organization, attendance is really essential. This process will be more inefficient and more time consuming if it is not managed smartly and using modern technology. In educational institutes, it is intricate to use the traditional approach of calling students names and maintaining attendance records when the number of students is high. Various methods are used by organizations to mark attendance like document-oriented approach, Card swipe, Biometric fingerprint, etc. In case if the card is lost or if the student forgot the card then the student will be marked as absent. Also, the student has to wait in the queue for this process. To overcome these drawbacks there should be a robust and reliable system to take the attendance of employees or students. Face recognition will be the more reliable approach for taking attendance. Face recognition does not necessitate a person's active participation. The primary contributors to the development of facial recognition systems are pattern recognition, face analysis, machine learning, and deep learning. The face recognition using CNN algorithm is more efficient and reliable. Fig -1: Basic block diagram of CNN The data is supplied to the model in the input layer. In the input layer, the number of neurons is equal to the number of features. There might be several hidden layers depending on the model and data amount. Each hidden layer has a distinct number of neurons. The hidden layer's output is supplied into the logistic functions. Using the logistic function, the output of each Class is transformed into the probability score of each class. This paper compares various machine learning, deep learning, IoT based approaches for face recognition. This comparative analysis of several publications will be beneficial in the implementation of a sophisticated and intelligent attendance system. For feature extraction, transformation, and recognition, deep learning approaches employ a multi-tiered course in a hierarchy of processing units. The algorithm begins to create a statistical input with each encounter and learns until the outcome reaches an acceptable degree of accuracy. The use of these modern techniques for face recognition will make the attendance system more flexible and will also reduce human errors. 2. Literature work 2.1 Technical Survey Fig2 represents the Maximum face recognition rate using the Local Binary Pattern Histogram (LBPH) algorithm for single face, multiple faces, group faces.