Ask a question from expert

Ask now

Masked Face Detection: A literature review

4 Pages3512 Words124 Views
   

Added on  2020-10-01

About This Document

This literature review covers how machine learning is used to detect face masks during COVID-19 situations. It discusses datasets, face detection methods, and masked face classification using CNNs.

Masked Face Detection: A literature review

   Added on 2020-10-01

BookmarkShareRelated Documents
Masked Face Detection: A literature reviewTrang H NguyenComputer Science Department, Earlham Collegethnguyen17@earlham.edu1.IntroductionIn 2019, Corona-virus disease severely affected the world.Ac-cording to the World Health Organization presented in WeeklyEpidemiological and Operational report on August 30th,2020[1]corona-virus disease 2019 (COVID-19)has caused over800,000 deaths and affected more than 25 million people fromatleast37 countries [2].There have also been some severediseases in the past few years named SARS (Severe acute respi-ratory syndrome) appearing in 2002 [3] or MERS (Middle Eastrespiratory syndrome) reported in 2012 [4] which also causedsimilar severe damage directly attributable to respiratory dis-eases. There is a report by Liu et al. [5] showed that the numberof COVID-19 cases is higher compared to the SARS. Hence, itcan easily to understand that the concern nowadays is people’shealth as more and more deaths keep occurring because of theseviruses. Governments from all over the world have been puttingpublic health as the top priority[6].Luckily, research shows that one of the simplest but effec-tive way which help prevents the spread of the virus is the sur-gical face mask [7].Most public services such as supermarkets[8], museums [9] and restaurants [10] have strict requirementsfor customers that if they want to use the services,they haveto wear masks [11].However,this does not mean that every-one agree or volunteer to wear a mask.Some people think thatwearing a mask affects their freedom, others claims that it is un-necessary because their belief show that the rules do not applyto them and they are in denial [12].This means it is hard tokeep track of the number of the people who do not wear maskespecially in public places.A Google Scholar search for the keywords ”face-mask de-tection” returns more than 18,000 results.This suggests thatthis problem is of widespread interest in the computer sciencefield.In conclusion, face mask detection has become a crucialtopic for the computer science vision aspect to help society.In this literature review, I will first cover the significant re-search on how object detection and image classification relatedto this idea and how this research can help detect whether theperson is wearing a mask. Secondly, I will present a set of meth-ods should be suggested to use that I have been researched.Fi-nally, I will conclude by suggesting and listing out the expecta-tions for this research’s possible future work.2.DatasetEvery researches use different datasets for several purposes topredict result(s).This section introduces and explains the rea-sons why I choose these datasets.There are going to be three sources of the dataset I am goingto use.Dataset from Kaggle website (Mask Dataset) [13] con-tains more than 800 images; Real-world Masked Face Recogni-tion Dataset (RMFRD) [14] contains more than 90,000 imagesand MAsked FAces dataset(MAFA) [15] contains more than30,000 images. All of the images in these datasets devided intothree type of images that are significant to identify the purposeof this research.Those images are:people who wear masks,people who do not wear masks,and people who do not wearmasks but occluded by other objects.There will be some cases that the machines might get con-fused, based on different situations such as the machines mightmiss-understand and detectsomething like a scarf as a mask(Figure 1).Each datasethas itown distribution,therefore Ichoose three datasets to evaluate my model.Although,threedatasets contain allthree type of images,each datasethas itsown unique and focus.The Mask Dataset mainly focus on thefirst two types: people who wear masks and people who do notwear a mask. The RMFRD dataset contains all three types: peo-ple who wear masks, people who do not wear masks, and peo-ple who do not wear masks but faces were occluded by otherobjects.The MAFA dataset also contains all three types of im-ages.More interesting, this dataset has many images with dif-ferent face-angles,such as right-side faces and left-side faces.For this research, I will mainly use the MAFA dataset and theRMFRD dataset. The Mask Dataset can be used to measure howwell our model generalize to new data as an optional extensionto the main project if time permits.2.1.Mask DatasetThis dataset[13] contains 853 images belonging to the threetypes: with mask, without mask, mask work incorrectly, as wellas their bounding boxes in the PASCAL VOC format.2.2.Real-world Masked Face Recognition DatasetThe authors of this research [14] proposed masked face datasetsin their research - Masked Face Detection Dataset(MFDD),Real-world Masked Face Recognition Dataset(RMFRD) andSimulated Masked Face Recognition Dataset (SMFRD). How-ever,we willonly discuss the RMFRD dataset,since itthedatasetthatreflects images of realpeople in this world.Theauthors also stated thatthis datasetis the real-world biggestmasked dataset that is free accessible.According to the authors of this dataset,they selected im-ages of celebrities and well-known people. After that they usedPython-crawled tool to crawled and cropped the front-face ofthose images.With some celebrities and well-known peoplethat they cannot find or cannot access to the pictures they weremasks,the authors took images from Internet then correspondthe images into simulated mask-face images by putted the mask- images in the face as pretend those images contain person wearmasks. In conclusion, the dataset includes 5,000 pictures of 525people wearing masks, and 90,000 images of the same 525 sub-ject and people without masks.2.3.MAFA DatasetThis dataset [16] was collected from a set of facial images fromthe Internet with more than 300K images from Flickr,Googleand Bing.The authors searched the images for the dataset withthe keyword “face; mask; occlusion and cover”.After that, the
Masked Face Detection: A literature review_1
Figure 1: Example images of Mask Dataset (1), MAFA Dataset (2), and RMFRD (3,4)author eliminated images that only contain faces but lack occlu-sion;and then narrowed down to images with 80 pixels mini-mum side length.Hence, they ended up having 30,811 imagesand 35,806 masked faces.Some images in this datasethavemore than one masked face.There are six attributes that manu-ally annotated for each face: face location, eye location, maskslocation, face orientation, occlusion degree, and mask type.These data also contain various angles from a person’sface.Meaning there are images showed front-face angle,im-ages showed left-face angle and images showed right-face an-gle.This helps the machine become flexible since it would bemore challenging for left-side and right-side faces to detect.The masked image in this dataset also considers multipletypes of label such as simple mask (man-made objects with purecolor), complex mask (man-made objects with complex texturesor logos) and human body (face covered by hand, hair, etc.) andhybrid ask (combinations of at least two of the aforementionedmask types) but I think we might count this as human body asan occlusion.To sum up,this dataset is a challenging datasetfor face detection but it contains diverse face poses in the realworld.3.MethodsMasked face detection problems can be divided into two stages.The firststage is Face Detection - detection the faces in theimages ofdatasets.In this stage,we willuse HaarClassi-fier [17] and Cascaded ConvolutionalNetworks [18] method.Haar Classifier can easily apply for all of the three type of im-ages mentioned in section 2,however this method works bet-ter with front-faces angle.Thatis why I also consider usingCNN - which is works more effectively with both front-facesangle and other different side of faces angle (mentioned in theMAFA dataset in section 2 about having left-side faces anglesand right-side faces angles).The second stage is predict if thedetected faces are wearing masks or not.This section analysessome methods used by major researchers for each of the above-mentioned stages.While much research has been done on thefirst stage, the second stage has recently caught researchers’ at-tention recently.3.1.Face Detection3.1.1.Facial Feature Detection Using Haar ClassifiersBased on the Haar feature,Viola and Jones [19] proposed amethod called Haar Classifiers uses to detected objectin themost rapid way.This method can also uses to adapted in de-tect human faces. The common Haar Feature divided into threefeature steps:Edge feature,Line feature and Center-surroundfeature.These features run all over the images in order to de-tect 5 important parts of the face including: eyes (2 eyes), nose,mouth, and overall face shape. This would help in analyzed theareas of the images to the location with the highest probabilityof containing the feature.Although,the limitation of this fea-ture is it can only detect front-faces angle, the number of falsepositives reduced and the detection speed were increased.3.1.2.JointFace Detection and Alignmentusing Multi-taskCascaded Convolutional NetworksThe author of this paper [18] proposed a a deep cascaded multi-task framework for face detection and alignment in order to helpdetect faces in different and difficult environments such as facesthat had various poses (include left-side faces angle and right-side faces angle) or faces under illumination and occlusion con-ditions.The method contains three main importantstages topredict face and landmark location: Proposal Network (P-Net),Refinement Network (R-Net), Output Network (O-Net).To bemore specific, when the machine receive a picture as an input,firstly,it is going to test the images by resize the picture intoimage-pyramid ( meaning divided the same picture into dif-ferent sizes ) so that the machine can detect the picture morespecifically under differentdetails.Nextup,the image willbeing process under P-Net step,where the output of this stepwould release boxes to square that marked the overall face inthe image. Meaning the machine going to ”scan” and try to findmultiple boxes that it consider is the human-face, then eliminatesome boxes with less confidence (since some boxes smaller thanthe others,which the percentage is definitely lower).The rea-sons why this stage contain more than one box although some-times picture only contain one face is because in some cases,picture contains more than one face or have more than one per-son.After thatis the R-Netstage,which is pretty similar toP-Net stage.However this stage would keep eliminate boxesthat have lower confident of face-detected,and more focus indetect the shape and the angle of the face in the image so thatit can define the final box which fit with the face in the image.This would lead to the final stage - O-Net.Base on the box ofthe face, this stage finalize the position of the 5 most importantplaces in the face: 2 eyes, 1 nose and 2 corners of the mouth.Throughout this paper, we can have a bigger picture of whatface detection is, the difference, and how this method can helpdetect a person’s face which can help a lot in using for challeng-ing dataset like MAFA.3.2.Masked Face Classification3.2.1.Masked Face Classification Using Convolutional NeuralNetworkConvolutionalNerualNetwork has been widely used for Im-age classification tasks,based on structure of LeNet [20] andAlexnet [21]. They have applications in image and video recog-nition, recommendation systems, image classification, medicalimage analysis, natural language processing and financial time
Masked Face Detection: A literature review_2

End of preview

Want to access all the pages? Upload your documents or become a member.

Related Documents
COVID-19 and Tourism: Impact, Challenges, and Recovery
|9
|1966
|476

The disease was formally named Corona Virus
|7
|1672
|11

Coronavirus: Relevant Facts
|4
|827
|11

SARS Outbreak in China
|13
|3873
|24

Essay on COVID 19
|7
|1992
|165

Public Health Outbreak - Novel Coronavirus
|4
|746
|14