Covid symptom detection + Mask detection

Abstract

– Head pose classification is widely used forthe preprocessing before face recognition and multiangleproblems,becausealgorithmssuchasfacerecognition often require the input image to be a frontface. But affected by the COVID-19 pandemic, peoplewearfacemaskstoprotectthemselvessafe,whichmakes cover most areas of the face. This makes somecommonalgorithms cannotbeappliedto headposeclassificationinthenewsituation.Therefore,thisproject established a method HGL to deal with theheadpose classification by adopting color texture analysis ofimages and line portrait.Theproposed HGL methodcombines the H channel oftheHSV color space withthefaceportraitandgrayscaleimage,andtraintheCNNto extract features for classificationand its hardware pulsesensorchecktheoxygenlevel of personandnotifyandalsocheckTemperaturebyusingtemperatureSensor. Notify the level of Temperature low or not. Theevaluation on MAFAdataset shows that compared withthe algorithms based on facial landmark detection andconvolutional neural network, the proposed method hasachievedabetterperformance.