Stellenbosch University
Welcome to Stellenbosch University
Top computer science students recognised for 2020 achievements
Author: Media & Communication, Faculty of Science
Published: 29/11/2021

​Working on improving the automated image recognition of individual leopards from photographs for her BScHonours project in Computer Sciences at Stellenbosch University, Ms Jacobie Mouton not only published her first peer-reviewed article in an academic journal. She has also received the award for the best BSc (Hons) student in Computer Science for 2020, as well as the award for the best BSc (Hons) project in Computer Science. 

During an awards ceremony on Tuesday 24 November, Prof Bernd Fischer, head of the Computer Science Division in the Faculty of Science, personally congratulated each of the top performing computer science students from last year. They are Bernardus Wessels (best first year student), Brendan Watling (best second year student), Matthew Baas (best student in machine learning), and Caleb Zeeman (best third year student and recipient of the Van der Walt medal).

ComputerScience topstudents2020_banner.png

Prof Fischer also thanked the division's sponsors for their support for the department and generous prize monies for the students. These companies are Epi-USE, part of one of the world's largest independent SAP HR/Payroll specialists (; the software engineering and electronics company Alphawave, the machine learning company Praelexis, telecommunications company VASTech, and the accounting technology company Techairos.

More about Jacobie's research

For her BScHons-project, Jacobie developed a new and much faster method for the automated extraction of the regions of interest (ROI) from photographs of leopards, which is used for the identification of individuals based on the rosette patterns contained in the ROIs. In automated image recognition, the correct identification of the region of interest helps the  identification algorithm to more accurately match a photograph to a database of known individuals by eliminating the irrelevant parts of the image that contain background objects . In the case of images from a camera trap, for example, the computer program needs to be able to differentiate the rosette pattern of a leopard from the surrounding trees, rocks and grass. Currently, this process is performed manually.

In order to develop and test this new method, she worked with a database of images from the Kgalagadi Leopard Project, containing 260 images of 40 known individual leopards taken from various angles.

She says automated identification can save researchers a lot of time when having to process large volumes of data: “Since camera traps can produce great volumes of images, manual She says automated identification can save researchers a lot of time when having to process large volumes of data: “Since camera traps can produce great volumes of images, manual segmentation is labour intensive and time consuming for researchers. Automated segmentation is therefore a valuable tool that can be utilised to great effect".

For a proof of concept of the new method, visit her project at

In her academic career thus far, this former learner from Hoër Meisieskool Bloemhof has already received a number of awards, including the Van der Walt medal as the best final year student in Computer Science. In 2020 she became part of the first cohort of MSc students receiving a DeepMind scholarship for postgraduate studies in artificial intelligence and machine learning.

On the photo above, Jacobie Mouton. Photo credit: Stefan Els