Detection system of facial patterns with masks in new normal based on the Viola Jones method

Abstract

Covid-19 is something that was never expected, it can turn into an endemic virus in the community. There is a possibility that this virus will not be completely destroyed. This makes the world and Indonesia in an uncomfortable position. Two months with Social Distancing conditions, the Government of Indonesia has been preparing to roll back the sluggish economic wheel as a result of the implementation of Social Distancing. Therefore, the Indonesian people must live in peace with Covid-19 until the discovery of an effective vaccine. This condition is called new normal. This study designed a detection system of facial patterns using masks during the pandemic based on Real-Time Raspberry. The purpose of detecting face patterns by using a mask is to find out if there are masked faces in the image. Although it seems easy to do by humans, it turns out that this detection system is difficult to do without the help of a computer to process facial recognition because there are some difficulties related to location, point of view, light, and occlusion. This research has implemented a detection system using the Viola Jones method. Viola Jones method is a method to get fast, accurate and efficient results in face detection on images. This study using the Viola Jones method to adjust the threshold value, and form the Cascade Classifier in determining the face area in the image. This training can be evaluated the accuracy of the system by modifying the parameter values in the Viola Jones method so that this design can produce the highest accuracy for face images using masks and low accuracy for face images without using masks. From the results of trials with 100 face samples, the accuracy percentage is 90.9% and it takes a relatively short time to detect faces using a mask that is on average 15 seconds per sample tested.

Incorrect Facemask-Wearing Detection Using Convolutional Neural Networks with Transfer Learning

“…On the one hand, references [18][19][20][21][22][23][24][25][26][27] developed a mask detection system that can differentiate between the presence or absence of a mask. On the other hand, references [29][30][31][32] are also able to detect if the nose is outside the mask.…”

“…In a study carried out by Jauhari et al, in [21], the aim was to detect facial patterns to be able to detect the presence of facial masks in images. For this, it was based on Single Board Computer (SBC) Raspberry Pi.…”

Face Mask Detection in Smart Cities Using Deep and Transfer Learning: Lessons Learned from the COVID-19 Pandemic

“…Works that build on face detection and then extract image features to perform FMD employ decision tree classifiers. These give a speedy, simple, and efficient method for detecting masks [51]. Similarly, other techniques build on conventional object detection approaches, such as Haar Cascade [52] and Histogram Of Gradients [53] to detect valuable features, which are then employed by an ML classifier to solve the FMD task.…”

Applying Gamma and Histogram Equalization Algorithms for Improving System-performance of Face Recognition-based CNN

“…One of the machine learning algorithms known for its power in detecting faces is Viola-Jones [5]- [7]. It consists of four main stages which are: Integral Image, Haar Features, Haar Feature Classifier, and Cascade [8]- [10]. One of the most important problems that strongly affect the accuracy of the results of the viola algorithm is a low-quality image, especially in terms of lighting, if the image is very bright or dark [11], [12].…”