Motion detection and Face Recognition to detect people wearing masks in preventing the
transmission of covid 19 using the Haar Cascade Algorithm
Dedy Rahman Prehantoa, Aries DwiIndriyantia, TerdyKistoferb
a
Department of Informatics Engineering, Universitas Negeri Surabaya, Surabaya, Indonesia
bDepartement of Information System, Faculty of Information Technology, Universitas Hasyim Asy’ariTebuireng, Jombang -
Indonesia
Article History: Do not touch during review process(xxxx)
_____________________________________________________________________________________________________ Abstract: The development of technology in biometrics is currently growing rapidly, especially facial recognition. Face detection is used to detect the location of the face in doing recognition. In face detection research, many different algorithms are used to maximize the detection results. The algorithm is used to improve accuracy and processing speed. Because currently Indonesia is still affected by the corona virus 19 (Covid 19) so it is mandatory for residents to wear masks wherever they are and apply health protocols. So this study develops a Motion detection and Face Recognition to detect people wearing masks. in preventing the transmission of covid 19 using the Haar Cascade Algorithm. The Haar Cas cade Classifier algorithm is an algorithm that is widely studied because it can provide unparalleled precision for object detection. From this research, the accuracy value of 99% is obtained. With good lighting, good photos will give you the best results, several factors such as minimum exposure, optimal shooting angle and distance factor when shooting photos of different sizes so that the data sent can be optimized.
Keywords: Face recognition, mask, haarcascade algorithm, Motion detection
___________________________________________________________________________
1. IntroductionThis Since being designated as human-to-human transmission in Wuhan, China on December 31, 2019, infection with the coronavirus-2019 (COVID-19) that causes acute respiratory syndrome-Coronavirus 2 (SARS-Cov-2) has become something of a global pandemic (Ridlo, 2020).Meanwhile, in Indonesia, only about 3 months after the government announced the first two confirmed cases of COVID-19 in the Indonesian capital, the number suddenly rose to 45,891, the highest in Southeast Asia ( Wordmeter, 2020). Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) is a type of corona virus that causes COVID-19 and has spread in all 34 provinces in Indonesia (World Health Organization (WHO) Indonesia, 2020). In terms of death toll, more than 2,000 people in Indonesia have died from SARS-CoV-2. This number may be much higher as the country may still be reporting the number of asymptomatic infections (World Health Organization (WHO) Indonesia, 2020). In addition, testing rates in Indonesia, the world's fourth most populous country, are still low. As of June 22, Indonesia had conducted 639,385 tests on 273 million people, or about 2,338 tests per million people (Worldmeter,2020).
The number of COVID-19 cases in Indonesia continues to increase, and as of March 22, 2020, the number of positive COVID-19 cases was 514, with 29 recovering (5.64%) and 48 dying. Asia. In this world, a disease that plagues and causes death for which there is no cure (World Health Organization (WHO) Indonesia, 2020). The trend map and distribution of the number of COVID-19 cases in Indonesia is presented in Figure 1 and 2.
Figure 1. Trend of the number of positive cases for COVID-19 in Indonesia on March 2020 Negative Positive(World Health Organization (WHO) Indonesia, 2020).
Figure 2. Spread map of COVID-19 in Indonesia on March 2020 (vina,2020)
Clinical symptoms of this viral infection range from runny nose (fever, cough, runny nose, sore throat, muscle aches, headache) to serious life-threatening complications (diarrhea and pneumonia)(Huang et al, 2020). The increasing spread ofCOVID-19 is due to a lack of public awareness of compliance with health
protocols.Therefore, in times of the coronavirus pandemic, with anticipated infections and increasing numbers, it is advisable to adopt a new healthy lifestyle according to public health protocols ( Purwaningrum et al, 2021). One form of the procedure is to maintain cleanliness and avoid direct contact with patients who are positive for the corona virus (Izzaty,2020). Then wash your hands or use hand sanitizer to keep themclean (Lee et al, 2020). Then use a protective mask when traveling or leaving the house (Howarda, 2020).
Based on the problems that exist in the community, we will take advantage of the development of multimedia technology, the use of cameras and video/image processing to prevent the transmission of the corona virus. One of the camera and image/video processing applications is facial recognition. Security cameras (Peter,2011), attendance (wagh, 2015), smartphone security applications (Kremic,2012) and used for online games (Zhan, 2008) Some applications recognize faces (face recognition). So this study develops a Motion detection and Face Recognition to detect people wearing masks. in preventing the transmission of covid 19 using the Haar Cascade Algorithm. The Haar Cascade Classifier algorithm is an algorithm that is widely studied because it can provide unparalleled precision for object detection (Sanjaya, 2017). which is one of the methods of machine learning (Prehanto,2021). With this technological system as an effort to prevent the corona virus by complying with health protocols among the public, especially when entering a room in a public place
2. Research Methods
Steps to use HaarCascadeAlgorithm to design a system to detect who is wearing a mask to prevent covid19 infection The following Figure 3.
2.1. System analysis.
This system is built with the Python programming language using the OpenCV library. Face imaging is done directly using a webcam camera connected to a computer/laptop to be used as a dataset/training data. The image function uses the haar layer algorithm. Arduino control data by microprocessor opening and closing door portal.
2.2. HarcascadeAlgoritm.
Object discovery using the Cascade Classifier-based Haar function is an object discovery method. Paul Viola and Michael Jones (2001) published a paper entitled "Fast Object Detection Using Simple Extensions". A Haar layer is a collection of Haar type functions that are combined to form a classifier. The attribute is the sum of the pixel values in the black area minus the white pixel values. The Haar like function, also known as the Haar cascade classifier, is a rectangular function that provides a specific representation of an image. The Haar Cascade classifier is derived from the combination of black and white pixels that make up the mesh (Rahim, 2013).
The mask detection process uses the Haar Cascade algorithm. Features such as haar are usually used to detect objects from digital images. The word Haar itself is represented as a mathematical function in the form of a box (wavelet Hhaar). Initially, image processing was based solely on the RGB value of each pixel, but this method was inefficient. Viola and Jones then developed and designed the Haar-Like feature. The haar style function processes images in a grid with multiple pixels in a single image. It then processes each rectangle, generates another value, and displays the light and dark areas. This value is used as the basis for image processing (Pavani,2010).
Figure 4. Haar Like Features (Pavani, 2010). 2.3. Cascade klasification.
A tiered classifier is an iterative calculation of the value of a Haar object for more precise results. Figure 2 shows the stratified classifier workflow. In the classification stage 1, each sub-face is categorized according to its characteristics, and if the results do not meet the requirements, the results are rejected. In the two -step classification, each soffit is reclassified. When the required threshold is reached, the next screening step (rank step 3 begins). Until the skipped sub-image is reduced to close to the sample image being tested. The cascade classifier consists of stages, including a powerful sorter. Therefore, all functions are divided into several stages, and each stage has several functions. The work at each stage determines whether it is a given window object. If the process does not respond to the same attribute value, it will indicate that there is no object. The author uses the OpenCV Haarcascade library for the detection process. Each level has several features. The actions at each step are used to determine whether a particular panel is an object. If the process does not respond to the same attribute value, it will be marked as non-object. The author uses the OpenCV Haarcascade library for the detection process (Zhao and Wei, 2017).
2.4. OpenCV
OpenCV (Open Source Computer Vision Library) is a computer vision and machine learning (Opencv) library based on an open source project. It is produced by Intel, which specializes in image processing, whether image or video. OpenCV has more than 2,500 optimization algorithms, including classic and advanced machine learning and a comprehensive set of machine learning algorithms. This algorithm includes human or animal face detection and recognition, object type recognition, human behavior recognition in images in video or image format, camera change tracking, subject movement tracking, and others (Zhao and Wei, 2017).
2.5. Arduino microcontroller as a door open and close control.
Arduino Uno is one of the products labeled Arduino which is actually an electronic board that contains the Atmega328 microcontroller. The board has 14 digital input/output pins (6 of which can be used as PWM outputs), 6 analog inputs, a 16 MHz crystal oscillator, a USB connection, a power jack from the reset button. This pin contains everything needed to support the microcontroller. You can connect it to your computer with a USB cable, or get a power source from an AC-DC adapter or battery. Arduino to the microprocessor runs the motor control process by opening and closing the gate when it detects someone wearing a mask (Lestari, 2017). 3. Results And Discussion
The first step in performing face recognition is a training program so that each image can store a different face definition sorter for 30 human face images with different RGB, from dark to light areas. The resulting image has a .jpg extension and a resolution of 640x480 pixels or less as follows:
Figure 6. Example data set on user id 1.
The second step Next is to choose the id of the image to be detected so that it can be learned that id = 1 ; id = 2; id= to n. to recognize faces using trainer.yml on Open CV using the har cascade method with the values obtained will be processed using the haar like feature by calculating the difference from the number of pixels (Puspaningrum and saputra, 2018).The third step is face recognition recognition according to the id in the data set with the percentage accuracy of the name according to the id in the data set. The size of thispercentage is influenced by the level of lighting, facing the face forward, and how close it is to the camera. seen in the following image shows when one of the ids in the data set wears a mask or not:
Figure 7 . face detection test.
The next step is sending data to the Arduino so that the Arduino microcontroller can open or close the portal door automatically based on people wearing masks or not wearing masks. If a person wearing a mask is detected, the portal will open but not wearing a mask the portal will be closed. The following figure shows the results of the Arduino control testing system. The next step is sending data to the Arduino so that the Arduino microcontroller can open or close the portal door automatically based on the person wearing the mask. or not wearing a mask. If a person is detected wearing a mask, the portal will open but not wearing a mask, the portal will be closed. The following Figure 8 shows the results of the Arduino control testing system.
3.1. Hardware component testing.
Automated portal prototypes work in only one direction. The door can be opened after activating the first sensor and closing after activating the second sensor. Hardware Component Testing Hardware component testing uses ultrasonic sensors to look for defects in this prototype. Especially for the hardware components used to build the system.
3.2.Ultrasonic sensor testing.
Ultrasonic sensor testing involves the collection of parameters related to the accuracy of the distance detected by the sensor.
3.3. Servo Motor Test.
This servo test aims to see how efficiently the servo moves (in degrees). The movement is activated by the servo wheel. Servo testing is performed at an angle of 180 degrees from 0 degrees to 180 degrees in the initial state Each angle is tested 3 times and the time is in seconds.
3.4.Testing system.
Table 1. implementation of the Motion detection and Face Recognition system by opening and closing the automatic door portal to detect people wearing masks in preventing the transmission of covid 19.
No Testing Face detection Result
1. Face Not
detected
Open Portal
2. Face
detected at
id=1 Closed portal
3. Face detected at id=2 ClosedPortal 4. Face detected without name. Closed Portal
Based on the test results of the Motion detection and Face Recognition system to detect people wearing masks that are connected to Arduino as an automatic door portal control, the success rate is the same in experiments 1-4 of 99%. Experiment 2 has an accuracy of 23% face detection rate according to id. And experiment 3 has a face detection accuracy rate of 39%. While in the experiment to id it is not detected but is detected not using a mask then the portal is closed. With good lighting, a good photo will give you the best results, as it will bring the face of the subject forward. Detection errors are sometimes caused by factors such as minimum exposure, optimal shooting angle and distance factor when shooting photos of different sizes. In addition, some images have small graininess caused by noise, especially the effect of night shooting.
4. Conclusion
conclusions Based on the discussion that has been carried out, as follows:
4.1.The ultrasonic sensor detects the arrival of people which is integrated with the Arduino UNO microcontroller; while the web cam detects people wearing masks or not so that they can be opened and closed automatically.
4.2.This harcas cade algorithm can be applied as a support system to determine the automatic door portal that will be inputted.
4.3.Motion detection and Face Recognition system to detect people wearing masks that are connected to Arduino as an automatic door portal control can be applied with a success accuracy value of 99% With good lighting, good photos will give you the best results, several factors such as minimum exposure, optimal
shooting angle and distance factor when shooting photos of different sizes so that the data sent can be optimized.
References
Howarda, Jeremy c, Austin Huangb ,ZhiyuanLik , Zeynep Tufekcim, Vladimir Zdimale , Helene-Mari van der Westhuizenf,g , Arne von Delfto,g , Amy Pricen , Lex Fridmand , Lei-Han Tangi,j , Viola Tangl , Gregory L. Watsonh , Christina E. Baxs , ReshamaShaikhq , Frederik Questierr , Danny Hernandezp , Larry F. Chun , Christina M. Ramirezh , and Anne W. Rimoint(2020) Face Masks Against COVID-19: An Evidence Review. Journal reference: PNAS.
Huang, C., Wang, Y., Li, X., Ren, L., Zhao, J., Hu, Y., & Cao, B. (2020). Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China. The lancet, 395(10223), 497-506.
Izzaty. (2020). KebijakanPemerintahdalamMengatasi Panic Buying Akibat COVID-19. Info Singkat, 12(1), 20–30.
Kremic,E., A. Subasi and K. Hajdarevic (2012) “Face recognition implementation for client server mobile application using PCA,” Proceedings of the ITI 2012 34th International Conference on Information Technology Interfaces, Cavtat, 2012, pp. 435 -440. doi: 10.2498/iti.2012.0455.
Lee, J., Jing, J., Yi, T. P., Bose, R. J. C., Mccarthy, J. R., Tharmalingam, N., &Madheswaran, T. (2020). Hand Sanitizers : A Review o n Formulation Aspects , Adverse E ff ects , and Regulations. International Journal of Environmental Research and Public Health, 17, 2– 17.
Lestari, N. (2017). “RancangBangunPintuOtomatisMenggunakan Arduino Uno dan PIR (Passive Infra Red) Sensor di SMP Negeri SimpangSemambang”. Jusikom, Vol 2 (2) .62-68.
Pavani, Sri Kaushik, David Delgado, and Alejandro F. Frangi(2010) “Haar-like Features with Optimally Weighted Rectangles for Rapid Object Detection.” Pattern Recognition.
Peter,K. J. ., G. G. S. Glory, S. Arguman, G. Nagarajan, V. V. S. Devi and K. S. Kannan (2011) “Improving ATM security via face recognition,” 2011 3rd International Conference on Electronics Computer Technology, Kanyakumari, pp. 373-376.
Prehanto, Dedy Rahman; Indriyanti, Aries Dwi; Prismana, I. GustiLanang Eka; Permadi, GinanjarSetyo; Prastyo, Edwin Hari Agus ( 2021) Implementation of Web Scraping on News Sites Using the Supervised Learning Method.journal of elementary education online (EEO);20(3):432-441; 39:449–52
Purwaningrum, ratna, Alif RizkyHafizhdillah ,AkhmadKheru , DwiRobbiardyEksa , Festy LadyaniMustofa , RakhmiRafie(2021) PenyuluhanPencegahanPenyebaran Virus Corona DenganMematuhiProtokol Kesehatan Ratna 1 , , [JURNAL KREATIVITAS PENGABDIAN KEPADA MASYARAKAT (PKM), P-ISSN: 2615-0921 E-ISSN: 2622-6030 VOLUME 4 NOMOR 1 HAL 200-206. Puspaningrum, Eva Y , Wahyu S.J. Saputra (2018) DETEKSI WAJAH DENGAN BOOSTED CASC ADE CLASSIFIER. ISSN : 1978-0087 -
SCAN VOL. XIII NOMOR 3 - OKTOBER 2018.
Rahim, Abdur, N. Hossain, T. Wahid, and S. Azam(2013) “Face Recognition Using Local Binary Patterns (LBP).” Global Journal of Computer Sience and Technology Graphics & Vision
Ridlo ,IlhamAkhsanu (2020)PANDEMI COVID-19 DAN TANTANGAN KEBIJAKAN KESEHATAN MENTAL DI INDONESIA.insanjurnal dan psikologi dan Kesehatan mental. vol 5.no 2.
Sanjaya, KadekOki,.GedeIndrawan, KadekYotaErnandaAryanto. (2017). PendeteksianObjekRokok Pada Video BerbasisPengolahan Citra DenganMenggunakanMetodeHaar Cascade Classifier. Journal of Natural Science and Engineering. Vol.1 (3) pp. 92-99.
Syarif, Muhammad, and Wijanarto( 2015) “DeteksiKedipan Mata DenganHaar Cascade Classifier Dan Contour Untuk Password Login.” Techno.com 14(4): 242–49.
Vina, M.F. (2020) 331.273 Orang Recovered, 97.847 Orang Sembuh", https: //www.kompas.com/tren/read/2020/03/23/072649465/update-virus-coronadi-dunia-331273-orang-Recovered-97847-orang-sembuh?page=2. [Accessed, 23rd, March 2020].
Viola, Paul, and Michael Jones( 2001) “Rapid Object Detection Using a Boosted Cascade of Simple Features.” In Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition.
Wagh,P. , R. Thakare, J. Chaudhari and S. Patil,(2015) “Attendance sistem based on face recognition using eigen face and PCA algorithms,” 2015 International Conference on Green Computing and Internet of Things (ICGCIoT), Noida pp. 303-308.
World Health Organization (WHO) Indonesia, Coronavirus Disease 2019 (COVID-19) WHO Indonesia Situation Report, Indones. Situat. Reports. (2020). https://www.who. int/indonesia/news/novel-coronavirus/situation-reports (accessed June23, 2020).
Worldometer, Reported cases and deaths by country, territory, or conveyance, (2020). https://www.worldometers.info/coronavirus/(accessed June22, 2020).
Zhan, C., Li, W., Ogunbona, P. &Safaei, F. (2008). A real-time facial expression recognition sistem for online games. International Journal of Computer Games Technology(Article No. 10), 1-7.
Zhao, Xue Mei, and Cheng Bing Wei (2017) “A Real-Time Face Recognition System Based on the Improved LBPH Algorithm.” In 2017 IEEE 2nd International Conference on Signal and Image Processing, ICSIP 2017.
