Taking attendance has been widely used as a method to track students' academic behaviours. But conventional approaches are usually time consuming and inaccurate. In this paper, we have proposed an automated attendance tracking system, where students can use smartphones to submit attendance in parallel. Identity of a student is collaboratively verified by voiceprint and realtime location. Service isolation and token-based mechanism is applied specifically in the system as well, for the purpose of protecting users' privacy and the system's security. Tests in real world environment has demonstrated that our system is characterised as high efficiency and accuracy in tracking attendance.

An automated student attendance tracking system based on voiceprint and location

This paper aims to implement a facial detection and recognition system through artificial neural network using a video recording as the input. The system will be used to generate a class attendance report that labels the status as either present, late, or absent to the correct student. The system will utilize the raspberry pi module to communicate with the pi camera, which takes the video recording of the class, this recording will be used as an input to the facial detection and recognition in order to determine the student statuses and eventually generate the class attendance. Confusion matrix will be used to measure and visualize the performance of the classifier and the algorithms. The researchers were able to obtain an accuracy of 87.70% for the attendance accuracy.

Class Attendance Generation through Multiple Facial Detection and Recognition Using Artificial Neural Network

This paper proposed a reliable recovery mechanism for person-following robot in case of missing target. The prior information such as previous human positons (before losing point) is used in order to predict the unexpected human positions by probabilistic approach (using Kalman Filter). In addition, map information is also utilized which allows robot to grasp searching route in order to redetect human target. Map also assists the robot in navigating smoothly with obstacle avoidance function to reach the human target. The experimental results demonstrate the outperformance of proposed recovery mechanism.

The reliable recovery mechanism for person-following robot in case of missing target

Tracking people even being partially or fully occluded in the group situation is studied using a Perception Sensor Network (PSN) system which is composed of multiple Kinects used to detect human 3D locations, and pan tilt zoom (PTZ) cameras used to identify human faces. A method is proposed to fuse multiple detection of human in the PSN system. After associating detected human with corresponding names, the novel grouping and ungrouping algorithms are proposed. When a group of multiple human staying close together is formed, viewpoint and illumination invariant features of group members including human 3D location, height, color and binary robust invariant scalable keypoint (BRISK), retrieved from region of interest (ROI) of both depth and color images, are then stored and updated into the group database. Based on the distance between a group location at previous frame and each member location in the group at current frame, the PSN system decides whether to keep the members in the group or to ungroup them then reassign the right name among the group database by minimizing multiple criterions. The experimental results demonstrate the outperforming of the proposed method on tracking people in group than conventional methods.

Robust Tracking Occluded Human in Group by Perception Sensors Network System

The university system in the Kingdom of Saudi Arabia is concerned with student attendance for lectures, and it is the responsibility of lecturers to monitor student attendance for each lecture. By the end of the semester, students get an attendance register indicating which lectures the student has attended and it reports the calculated percentage for each student’s attendance in each course. Universities have regulated the mechanisms and the acceptable percentages of student absence. The process for a lecturer to manually check student attendance consumes a lot of time and effort, either during the lecture or when in the process of emptying absenteeism and inserting it into the university’s electronic system. Therefore, Saudi universities compete to find modern methods of checking student attendance that will avoid the disadvantages of manually taking attendance. For this reason, they have produced electronic attendance systems, for example, using a student's fingerprint, an eye recognition system, or a mobile phone system to read a QR code designed for the same purpose. All of these systems have the disadvantage that they consume a lot of time, as all students have to line up at the fingerprint reader or the eye detector for identification. Therefore, the problem of the consumption of lecture time is still present, even with these modern systems. Therefore, the aim of this research is to propose a smart mobile application that is able to check the attendance of students without having to consume lecture time or require any effort from the lecturer. The system automatically recognizes the attendance of students through their university ID cards. Each lecturer would use his/ her own mobile phone to use the proposed system to check the attendance of students instead of using manual method to register the attendance of students and the students’ ID cards that are detected by coming within range of the lecturer reader would represent present students, and missing student ID cards represent absent students

/pdf/monitoring-student-attendance-using-a-smart-system-at-taif-2axuius2u3.pdf

Monitoring Student Attendance Using A Smart System at Taif University

In this paper, we introduce a simulation approach for emulating a real-world human recognition system called a Perception Sensor Network (PSN). The proposed PSN system has fusion components for automatic localization and facial recognition that uses multiple Kinect sensors and pan-tilt-zoom (PTZ) cameras. We verified that the generic vision schemes utilized in human detection and facial recognition algorithms were interoperable in a scenario that consisted of virtual avatar humans. In experiments, we tested a perception scenario in which multiple humans in a 3D simulation space are automatically recognized in both location and identification with real-world system parameters.

Realistic 3D simulation of multiple human recognition over Perception Sensor Network

This video demonstrates a multiple human tracking using multiple Kinects for checking the attendance of students in a smart class. We proposed sensor fusion mechanisms for seamless and consistent human tracking among Kinects. We evaluated the proposed system with six human in a classroom by checking whether human are located on their seat.

Multiple human tracking using multiple kinects for an attendance check system of a smart class

In this paper, we present a perception sensor network (PSN) system for automatically tracking location and identification of multiple human. We propose a fusion method for seamless human tracking over multiple kinects and for tracking human name using pan-tilt-zoom camera operations. By assistant with PSN, mobile robot understands multiple human' identification and location information while robot is moving. Finally, we have tested a robot scenario of dispensing paper materials to multiple human cooperating with PSN.

Dispensing materials of mobile robot cooperating with perception sensor network

The reliable perception of a human in a dynamic environment is the most critical issue for interactive human-robot services. In human-robot interaction, a camera on a robot naturally captures the low-body-part of human because robots are usually shorter than the human. Conventionally, a two-dimensional laser range finder is used in low-body-part detection [1, 2]. However, these methods may cause errors when there are similar structures with legs. This video demonstrates a low-body-part detection scheme that not only exploits three-dimensional characteristics and but also the RGB features of the low-body-part. We build the low-body-part candidates by clustering from the legs to the heap. In the results, spurious candidates are eliminated by the proposed method.

JiGwan Park

Papers

Realistic 3D simulation of multiple human recognition over Perception Sensor Network

Multiple human tracking using multiple kinects for an attendance check system of a smart class

Dispensing materials of mobile robot cooperating with perception sensor network

Low-Body-Part Detection using RGB-D camera