ABSTRACT
In this study, Convolution Neural Network (CNN)-based learning method, which is well-used in deep learning is applied to facemask detection from facial images of diverse persons wearing varied facemasks. During the covid-19 era, the wearing of facemasks has become important to curb the spread of the deadly virus and mandatory at times in some places. To ensure rules on wearing of facemasks are adhered to, face detection and recognition systems are being adopted to be able to detect masks worn on the faces of individuals. While good results have been obtained from some of such systems, the data set of images are not as diverse in terms of race and the types of facemasks detected are typically surgical masks. However, in real-world application, the society is made up of people of different races, age groups and wearing different kinds of facemasks. Therefore, there is room for improvement in the training, testing, and optimization of the detection systems by introducing a more diverse set of image inputs. The convolutional layer in this study is made up of numerous convolution kernels which are used to compute different feature maps for representations of the inputs. The model is evaluated by varying the image data and optimizing the hyperparameters for improved performance in facemask detection. Statistical Analysis performed to obtain Accuracy of 76.79% and area under the curve result of 0.8525 that demonstrates the capabilities of CNN. © 2021 IEEE.
ABSTRACT
The COVID-19 pandemic that has ravaged the world since December 2019 caused disruptions in the engineering education sector as students in African universities were unable to learn virtually. From two recent multi-language, multi-cultural pan-African online surveys to assess the impact on students, over 6,000 responses showed the twin constraints of irregular electric power supply and poor internet connectivity to effectively participate in virtual learning. This project is aimed at developing an affordable and reliable power and communication device for continuous online learning for engineering students, showcasing the strength in Africa's diversity through a collaborative, multinational, multicultural, multi-lingual and gender-sensitive platform to solve this identified global African Engineering Education challenge. A collapsible 100-watt solar photovoltaic module charging a set of lithium batteries via the charge controller was used to power a laptop computer, a mobile phone and a 5-watts bulb simultaneously through a Direct Current/Direct Current (DC/DC) converter. An embedded modem in the device provided the wireless network for internet connectivity. The initial prototypes produced weighed less than 7 kg, and preliminary performance tests showed that the gadget was able to charge up a laptop and two smartphones totaling 45.5WH from 0% to 100% while the remaining backpack state of charge remains 12.8V at 88% (that is 12% depth of discharge). The power supply and communication device for continuous online learning for African engineering students will not only bring engineering solution collaboration among hundreds of engineers, technologists, and technicians from the entire African continent, but will also boost entrepreneurial skills for many African engineering practitioners when fully commercialised. © 2021 IEEE.