Barend van der Merwe

I am a lecturer within the Department of Geography, Geoinformatics and Meteorology at the University of Pretoria. I have served in my current appointment since April 2012. Before this, I served as a Field Scientist for two overwintering Marion Island missions (M64 and M67 respectively) and the Summer Takeover Mission at SANAE IV (Antarctica) in 2011. I have also held multiple appointments as either a teaching assistant or a tutor. My research interests cover methodology where I have a particular interest in Artificial Intelligence (AI), computer vision, and physical computing. My Masters' dissertation showed how the analysis approach one employs (either inferential statistics or spatial analysis) can change the conclusions drawn when provided with the same data. Later, in my PhD thesis, I showed that a specific type of neural network known as a Convolutional Neural Network (CNN) can be trained to classify outlines of barchan dunes into usable morphometric categories. I am currently supervising a MA student working on the use of AI as a tool to manage plastic waste management. I am also co-supervising a PhD student whose project focuses on using AI to approximate discharge data from catchments that are not actively monitored. I have lectured modules in physical geography, human geography, cartography, environmental science, meteorology, and programming. For the majority of these courses, I created the necessary material and for one module I developed all the course material. Recently, I have been involved in teaching modules with enrollments of over 400 students and I was the only instructor for a course where more than 700 students were enrolled. During my career as an instructor, I have developed several novel computational solutions to assist me in my duties. These include a graphical user interface (based on Tkinter) linked to an SQLite database to more rapidly capture marks and minimise entry mistakes. I have also developed tools based on computer vision to automatically grade bubble choice (i.e. multiple choice questions) and to assist in the grading of map-based questions by superimposing the model answer over the student's submission. The former case increases security since it is no longer needed to make use of a third party to generate the bubble sheet and carry out the grading. It also allows for the layout of the bubble sheet to conform to any guidelines set by the employer. For the map-based questions, since the superimposition is carried out within a computational environment it increases the convenience of marking (since no scripts have to be carried around) and the software automatically compensates for minor differences in printer settings that modify the margins (thereby altering the dimensions of the map). I am continuously looking for more innovative ways in which I can use technology as a means to improve the quality of my work or make my work easier.