Offres de technologies

The University of Liège promotes the technologies developed by its researchers. Among these, new products, processes or services are ready for licensing.

An available technology is a technology for which ULiège has the exploitation rights and which has reached a sufficient degree of maturity to be transferred to one or more companies.

This transfer takes place through a license agreement managed by the company Gesval.

The technologies available in the field of Artificial Intelligence are listed below. Secured intellectual property is guaranteed for them. 

• HitNet, a neural network learning faster and augmenting learning data set :  HitNet was developed by Deliège Adrien, Anthony Cioppa and Prof. M. Van Droogenbroeck from the Telecommunications and Imaging Laboratory (ULiège, Montefiore Institute, Depart-ment of Electrical Engineering and Computer Science). HitNet is a redesign of a simple network to reach excellent performances where the output layer is replaced by a Hit-or-Miss (HoM) layer.
• ViBe, a powerful pixel-based technique that detects the background in video sequences : ViBe was developed by O. Barnich and Prof. M. Van Droogenbroeck from Montefiore Institute, Department of Electrical Engineering and Computer Science of the University of Liège. ViBe is a powerful pixel-based technique that detects the background in video sequences 
• Semantic Background Substraction, an efficient motion detection in video using objects recognition : Semantic Background Subtraction is a novel framework for motion detection in video sequences. It was developed by M. Braham, S. Piérard and Prof. M. Van Droogenbroeck from the Telecommunications and Imaging Laboratory from Montefiore Institute, Department of Electrical Enginee-ring and Computer Science of the University of Liège.
• ARTHuS (generic online distillation for image segmentation): Semantic segmentation can be regarded as a useful tool for global scene understanding in many areas, including sports, but has inherent difficulties, such as the need for pixel-wise annotated training data and the absence of well-performing real-time universal algorithms. To alleviate these issues, we sacrifice universality by developing a general method, named ARTHuS, that produces adaptive real-time game-specific networks for human segmentation in sports videos, without requiring any manual annotation. This is done by an online knowledge distillation process, in which a fast student network is trained to mimic the output of an existing slow but effective universal teacher network, while being periodically updated to adjust to the latest play conditions. As a result, ARTHuS allows to build highly effective real-time human segmentation networks that evolve through the match and that sometimes outperform their teacher. The usefulness of producing adaptive game-specific networks and their excellent performances are demonstrated quantitatively and qualitatively for soccer and basketball games.technique to build adaptive real-time match-specific networks for human segmentation, without requiring any manual annotation (best paper award in CVPR CVSports 2019 workshop). Source code (python, GNU Affero General Public License v3.0): https://github.com/cioppaanthony/online-distillation