Research interests

I started my research career with an interest in computer graphics and image processing. During my PhD studies I looked at applications of differential geometry in polygonal mesh processing and then moved towards computer vision, where I investigated and developed real-time, short range structured light scanning solutions for 3D reconstruction and 3D face recognition. When I joined the Vision and Learning research group at Stellenbosch my focus shifted more to the application of machine learning principles in computer vision. I have collaborated on a few projects with the Electronic Systems Lab where we develop computer vision techniques for autonomous navigation, as well as the CSIR's Modelling and Digital Sciences unit where we apply machine learning to various environmental domains. My current interests include deep learning for image and video analysis, the combination of vision and natural language, and visual knowledge representation. I am also one of the creators and coordinators of Maties Machine Learning, a discussion forum and interest group spanning all relevant disciplines at Stellenbosch University.

Selected projects

Here is a list of some of the research projects that I am or have been involved with. I am in the process of creating a project page for each, with more detailed information, so please check back soon.

deep learning approaches to object detection and classification in images and video
machine learning applied to climatological data
motion estimation on mobile devices
multiple target tracking
modelling uncertainty in occupancy grid maps
stereo vision based SLAM
uncalibrated multi-view reconstruction
real-time dense stereo
probabilistic shape-from-silhouette reconstruction
structured light scanning systems
curvature-driven mesh processing
surface parametrization

Current postgraduate students

Belinda Matebese  (PhD)  co-supervising with Mapundi Banda (UP) and Daniel Withey (CSIR)
Path planning for wheeled mobile robots using an optimal control approach

Samuel Ofosu Mensah  (PhD)  co-supervising with Bubacarr Bah
Detecting eye diseases using ultra-widefield fundus retina images

Gregory Newman  (MSc)  co-supervising with Ben Herbst
Deep learning for video classification

Nevesh Rajaram  (MSc)  co-supervising with Ben Herbst
Generative adversarial networks

Brendan Tighy  (MSc)  co-supervising with Francois Smit and Hardus Diedericks
The application of machine learning for the prediction of wind speed and direction in the Gordon's Bay area

Reem Elmahdi  (MSc)  co-supervising with Josefine Wilms (CSIR)
Machine learning for water quality analysis

Mulanga Sinyosi  (MSc)
Regional CNNs for object localization

Peter Thompson  (MSc)
Fine-grained image classification with limited training data

Shane Josias  (MSc)
Action prediction in video

Simba Nyatsanga  (MSc)
Deep sequence-to-sequence learning for automated video captioning

Jeanne Daniel  (MSc)
Weakly supervised knowledge acquisition for natural language modelling

Luyolo Magangane  (MSc)
Deep knowledge modelling

Previous students

Jaco Briers  (MSc, 2019)  co-supervised with Francois Smit
Data-driven river flow routing using deep learning: predicting flow along the lower Orange River, Southern Africa

Russell Kingwill  (MSc, 2019)
Evaluating the effectiveness of neural network techniques in the forecasting of South African basic fuel prices

Jacques Marais  (MSc, 2018)
Automated elephant detection and classification from aerial infrared and colour images using deep learning

Melise Steyn  (MSc, 2018)  co-supervised with Francois Smit and Josefine Wilms (CSIR)
Short-term stream flow forecasting and downstream gap infilling using machine learning techniques

Marco Grond  (MSc, 2017)  co-supervised with Ben Herbst
Text detection in natural images using convolutional neural networks

Simon Streicher  (MSc, 2016)  co-supervised with Johan du Preez
A probabilistic graphical model approach to solving the structure and motion problem

Gideon Zuurmond  (MSc, 2015)  co-supervised with Ben Herbst
Accurate camera position determination by means of moiré pattern analysis

Alwyn Burger  (MScEng, 2015)  co-supervised with Corné van Daalen
Occupancy grid mapping using stereo vision

Pieter Holtzhausen  (PhD, 2015)  co-supervised with Ben Herbst and Vladimir Crnojević (Novi Sad)
Video surveillance incorporating pan-tilt-zoom cameras

Lloyd Hughes  (MSc, 2014)
Enhancing mobile camera pose estimation through the inclusion of sensors

Pierre Joubert  (MSc, 2014)
Thermal and colour data fusion for people detection and tracking

Mogomotsi Keaikitse  (MSc, 2014)  co-supervised with Natasha Govender (CSIR)
Long-term tracking of multiple interacting pedestrians using a single camera

Simon Muller  (MSc, 2013)  co-supervised with Ben Herbst
Planar segmentation of range images

Wikus Brink  (MScEng, 2012)  co-supervised with Corné van Daalen
Stereo vision for simultaneous localization and mapping

Daniek Joubert  (MSc, 2012)  co-supervised with Ben Herbst
Adaptive occupancy grid mapping with measurement and pose uncertainty

Michael Burke  (MScEng, 2011)  co-supervised with Japie Engelbrecht and Kristiaan Schreve
Visual servo control for a human-following robot

Robbie Vos  (MSc, 2010)  co-supervised with Ben Herbst
3D position estimation of sports players through multi-view tracking

Francois Singels  (MSc, 2010)  co-supervised with Ben Herbst
Real-time stereo reconstruction using hierarchical dynamic programming and LULU filtering