Back to the previous page
Jérôme Treboux
Monday 27 February 2023 08:30

Jérôme Treboux is a well-known face at the institute and a good example of what the School of Management, its Business Information Technology studies and the institute of informatics can offer in terms of career development. After an apprenticeship as a computer scientist with a professional maturity in a banking IT company, he joined the Business Information Technology course at the HES-SO Valais-Wallis in 2009. At the end of his Bachelor's degree, he was hired by Professor Dominique Genoud as a research assistant and, in parallel, successfully completed a Master's degree. After 15 years at the School of Management, he will leave the institute with a PhD from the University of Fribourg on the analysis of aerial images using Machine Learning (a field of computer science often linked to artificial intelligence and allowing a computer to learn mathematical models based on processed data). Back from a trip to Australia to present his research work, he tells us about his career and his projects.

Studying at the HES-SO Valais-Wallis: the opportunity to study in the heart of the Alps

His family name is not from Valais and when asked why he chose to study in Valais, Jérôme makes no secret of his passion for mountain sports. A fan of sealskin, mountaineering, climbing, and trail running, our canton was the perfect playground for the young man who settled in Sierre, a town that has been a pioneer in computer science education and research for 30 years. "It's an ideal combination for reconciling private and professional life," he says. So Jérôme didn't hesitate at the end of his Bachelor's degree when Dominique Genoud suggested he join the DUDE-Lab (Data Understanding Data Explained) at the Sierre Institute of Computer Science. He began his research career there while continuing his studies with a Master's degree in Information Systems Management from the HES-SO.

Machine Learning: useful algorithms for many industrial fields

It was at the DUDE-Lab that the young researcher discovered Machine Learning, a sub-category of artificial intelligence that is based on the creation of information systems that learn from the data they process. The algorithm therefore improves each time it is provided with new data. This process makes it possible to create methods for solving problems and predicting results in fields as varied as tourism, medicine, agriculture, and energy. This is what Jérôme Treboux likes about his work. During his years of research, he has had the opportunity to discover many different fields of activity and new people, while at the same time broadening his field of expertise. And it is with this background that he embarked on a doctorate with the University of Fribourg on the detection of vine lines in aerial images taken by drone.

A doctoral thesis in the service of agriculture and viticulture

It is very important to Jérôme that all of his research work is applicable and helps to improve the daily life of the population. He therefore chose to write his doctoral thesis on an agricultural theme that is becoming increasingly important. Indeed, the topography of the Valais vineyards is steep and makes access difficult. In addition, the configuration of the vineyards is complex, with different orientations of the vine rows for each plot. To ensure economical and ecological phytosanitary treatment, drone treatment is increasingly used. But for this to work more autonomously, the detection of the vine rows by the drone's camera must be improved. Jérôme Treboux uses an innovative combination of Machine Learning algorithms that not only reduces the number of images needed for reliable detection, but also differentiates vine lines from other vegetation in the vicinity (e.g. a line of bushes or shrubs). Reliable detection of vegetation in aerial images is very complicated, especially because leaves change colour with the seasons and the orientation of the sun affects colours and shadows. But the advantages of automated detection brought by this project are numerous, since it will eventually allow the use of images from different types of cameras (satellite images or Swisstopo images, for example) with different resolutions, to apply fewer phytosanitary products by optimising the drone's route, to automate certain arduous tasks currently carried out by agricultural workers or to detect other types of vegetation.

Interest as far away as Australia - when Sierras projects go international

There are many wine-producing countries and Jérôme Treboux's doctoral thesis has found an echo as far away as Australia, which is one of the top five wine-producing countries in the world. As part of his PhD, Jérôme has published his research results which are presented in scientific journals and conferences. For example, he confirmed and shared the results of his research project at the IEEE 2022 Asia Pacific Conference on Computer Research and Data Engineering in Queensland, Australia. This conference on applied research allowed him to discuss with representatives from industry and academia and to make valuable contacts for the acquisition of future research projects for the HES-SO Valais-Wallis, in particular by presenting a little-known fact: Switzerland produces wine! Our researcher, who is thinking of moving to Australia for the rest of his professional career, will thus be able to maintain a privileged link between the Swiss and Australian research institutes. The presentation of our researcher's paper seduced the companies and universities who asked him to be a guest speaker at the 2023 Forest Grove Annual Conference. It is with interest and of course pride that the Computer Institute will continue to follow Jérôme Treboux's career on the other side of the globe.