Luis Fernando Cabarique Buitrago

L. F. Cabarique B. was a PhD student involved in the Underwater Perception and Sensing subproject of the Swedish Marine Robotics Center. The aim of his research was to develop fast, lightweight and robust algorithms for underwater place recognition. These algorithms were to be deployed in autonomous underwater vehicles (AUVs) with limited memory and processing capacity.

Since 2012 deep learning has been a dominant approach in the field of computer vision.  With a learning approach it is possible to separate training and deployment phases. Training of the model can take place in a device with significant computational power, while later deployment can be done in devices with limited computational capabilities. This allows optimization of computing resources and has the potential to reduce the time used to process an image. Because of these reasons Cabarique focused his research in the application of deep learning algorithms to the underwater place recognition problem.

Deep learning is a data driven technique and therefore compiling, cleaning and annotating data sets are a important part of the research process. During his studies Cabarique annotated and cleaned sonar and camera datasets, while also prototyping algorithms in Matlab and Python (Tensorflow and Numpy). Due to personal reasons Cabarique decided to retire from his studies and seek challenges in the industry.