Abstract—Today’s attendance marking are explicit, and timeconsuming in nature, they need at least some interaction withthe user. Manual entering of attendance in logbooks becomesa difficult task and it also wastes the time.Here an implicitattendance marking system is being proposed to mark theattendance of the employees automatically using face and motionrecognition, also saving a lot of time. This system does not requirea special attention of the user and it automatically marks theattendance as the employee walks through the gate. The use ofcognitive services, gives the user a message when their attendanceis marked. The system being implicit in nature does not needany special attention of the user.
The time being used daily intraditional systems is saved in this system. The register basedattendance markings quality of maintenance, and biometricattendance marking systems quality of explicit attention for thework is tried to improvise over here with the help of implicitnature of the system.The admin will be present here instead ofmany security guards, security guards, and other staff requiredfor this use. The user will need to initially register him/her inthe system, and thereafter the system recognises the employee onits own. The acknowledgement being sent to the user on theirpersonal device will be accompanied by the alerts of any meetings,or professional work along with personal greetings which mayinclude birthday/ anniversary wishes.
I. KEYWORDSInternet of Things, Embedded systems, face recognition,motion recognition, cognitive services, Camera.II. INTRODUCTIONThe system is designed for maintaining the attendance of theemployees present in the campus. Face detection is used fordetection of the person, and accordingly attendance is marked.In many campuses, the attendance is manually marked or ithas certain techniques like bio metric system and other traditionalattendance marking systems. We intend to make a smartsystem for marking attendance in which the person need notwaste his time standing in queue’s to enter his/her workplace.Also the effort of registering their face or themselves in thesystem can be saved.
here, The system will itself register aperson into the system by capturing several number of imagescropping, conversion, and analysis of the image data will bedone. The several landmarks and points on a person’s facewill be focused an the data will be stored. The landmarks, orthe distances between certain couple of points on a person’sface is unique just like someone’s fingerprint. The issue’sand challenges of being wrongly recognized by the systemcan also be recovered over here. Good quality of camerasand proper algorithms will help to store the data properlyand uniquely. The data as captured need not be stored andcompared as to save resources and avoid large computations.The data captured can be processed, analyzed and then storedin a compact format so as to increase the performance of thesystem.
The administrator can access the database, and also hasaccess to generate reports if any. The employee initially needsto register his/her details in a particular account which willalso store a number of images The face recognition will takeplace and user will get a message on their device regardingthe attendance.Emotions of the user will also be tracked and a monthlyreport will be generated for it. For this purpose, cognitiveservices can be used. User along with the acknowledgmentof attendance will also get certain important alerts, and alsogreetings, which consist of simple greeting and also if any importantpersonal event greeting like birthdays or anniversaries.The system being implicit saves time and other resources,and also important user efforts.
III. LITERATURE SURVEYTraditional attendance marking systems are explicit andtime consuming in nature, comparatively the automatic attendancemarking systems are implicit in nature and thereforedo not require special attention of the user, also they don’tneed to spent time on registration. The system automaticallyregisters the user and then processes the data to mark his/herattendance.Face recognition should be performed in real time to markdaily attendance of employees.
SMART-FR uses the advantageof handling multiple employees at the same time for markingattendance overcoming the drawbacks of previous algorithmsand techniques. The system is developed so as to operate underany conditions including bad light conditions and improperbackground. Important concepts such as Principal ComponentAnalysis (PCA), haarcascade are used. Handling leaves isgenerally performed in a manual way currently, using NaturalLanguage Processing or NLP, we can develop a system tohandle leave requests of the employee. 3Face recognition algorithms make use of Artificial NeuralNetworks(ANN) in the background to detect or to recognizethe face it already knows.
It compares the data with theprevious stored data, and analyses wheather it knows theface, it works as the human brain nerves to pass the inputinformation and process it. Feature vectors are extraceted fromthe human face, and a covarence column matrix is created, itis then converted into Eigen. Percentage calculater or analyzercan be formed so as to automatically calculate the percentageof attendance of a particular employee to monitor it afterwards.This can be used in companies or intitution having a certainconstraint or a threshold of minimum attendance. 4Human controlled or manually handled attendance can thenbe forged or manipulated, the automatic attendance markingsystems will thereby reduce or furthur eliminate the risk ofmanipulation with the records. Eigen Face matrix is calculataedand compared with the previous matrix and if the facevector is not present, a new face vector is added, and the tableor the data is updated with the additionof new value, else ifthe face vector matches, the attendance is marked.
1To avoid false detection, skin classification can be donebefore face detection and recognition. Skin clissification willstore the skin pixels and remove the colours fro the photograph,only the vectors in greyscale can be stored, and latercompared, so as to recognize the detected face. FFT and lowpass filters can be used for noise removal, and smootheningof images. 2IV. EXISTING ALGORITHMS Ada Boost Algorithm Float Boost Algorithm Principal Component Analysis Linear Discriment Analysis Active Apperance Model 3-D Face RecognitionV. PROPPOSED ARCHITECTUREThe proposed system in the below diagram consists of thefollowing modules and the interconnection between them:- Camera Module Sensor Module Raspberry Pi Module Controller Database Server Display devicesFig. 1.
System ArchitectureThe camera module has the basic function of capturing imageor video data. The data i.e.
images can either be captureddirectly or can be extracted from video. The real time dataprocessing consists of extracting faces from real time videocapturing, and then cropping the faces from captured video,and processing them. The sensor module consists of varioustypes of sensors including infrared sensors, hazardous sensors,proximity sensors, etc. the input which is obtained from boththe modules that is the camera and the sensor modules, isthen passed to the Raspberry Pi board for processing of theinformation.The Raspberry Pi board is the microcontroller we are usinghere for processing of the information. It is the componentwhich connects all the devices and is also responsible for theprocessing of the same. It is used to process the captured inputsand then pass them to the specific output devices, in order toimplement the function of the system.
If there is a new userof the system, he/she is initially registered into the system.The controller is responsible for the processing of queriesand the image data. The generated queries are passed to theapplication server for further processing. The File server isresponsible for the stored data of the existing Employees andto store the new registered employee data. The controllerdecides where to pass the data. The data on processing is againpassed to the Raspberry Pi board, to pass it to the displaydevices. The system records the face images, processes andstores the data related to it, and also stores the Employee idto which the data has to be mapped.
If the system recognizes acaptured face, it matches the data of the captured face with itsdata stored in its server and looks for the id of the Employeeto which it is mapped, after the extraction of the id, it marksthe attendance of the particular Employee. Here is the phasewhere the attendance of the particular Employee is marked,and the data is stored in the server for further requirements.Further, the attendance is sent to the display devices, whichmay be either the common display device present in thecampus, and also to the users personal device which mayinclude his/her laptop/mobile, etc.VI. ADVANTAGES AND DISADVATAGESThe flexibility in automatic attendance marking system ismore as compared to traditional attendance marking system.
Italso saves paperwork and eliminates errors such as duplicateentries. Also using Proxies for attendance is not possible.Real-time system is Implemented to deal with the real-timeenvironment. Automatic calculation of leaves and tracking ofemotions is aslo possible. Less efforts required for each typeof user. The System is implicit in nature. Automation makesthe system more easy to operate.Accuracy is not 100 percent efficient,the propsed systemmay have some limitations.
Initial investment is costly as comparedto tradiotional attendance marking systems.The overallchanges in the system such as intensity, background, changein look may prove as failures to the system sometimes.VII. CONCLUSION AND FUTURE SCOPEFrom the proposed system we can conclude that the drawbacksof the traditional attendance marking system are triedto recover and also an implicit approach towards marking ofattendance is brought forward. The system saves the precioustime of the employees spent daily using the other attendancemarking systems, the registration process is short enough andautomatic if a proper algorithm is implemented. System beingdigital, monthly reports can be generated.REFERENCES1 J. Joseph and K.
P.Zacharia. (2007, November) Automatic attendancemanagement system using face recognition.2 K. Selvi, P.
Chitrakala, and A. Jenitha. (2014, February) Face recognitionbased attendance marking system.3 J. Tha-Ranga, S.
M.S.C.Samarakoon, and T.A.P.Karunarathne. (2013,February) Smart attendance using real time face recognition.
4 J. otsana Kanti and A. Papola. (2014, June) Smart attendance using facerecognition with percentage analyzer.5 N. Kar, M.
K. Debbarma, A. Saha, and D. R. Pal.
(2012, Jully) Studyof implementing automated attendance system using face recognitiontechnique.