fr | en
Laboratoire Angevin de Recherche en Ingénierie des Systèmes

Séparés par des virgules

Séminaires des doctorants - Elie Karam11h00 | Polytech Angers | B106 | 62 Notre Dame du Lac - 49000 Angers

Sujet : Graph Neural Networks for Image Segmentation: An Approach with Physics-Informed Techniques and Multiview Perspectives.

Résumé

Graph Neural Networks (GNNs) have been seen to elevate semantic image segmentation by leveraging spatial and semantic relationships. We will explore the utility of "Informed neural networks" strategy in image segmentation. In this context, we propose to integrate domain-specific knowledge into the segmentation process, making them a versatile tool for combining knowledge-based insights with data-driven approaches in image analysis. This allows the model to capture both observed data and adhere to the fundamental principles of the underlying physical structure. We aim at exploring another approach, based on multiview models, that could help integrating heterogeneous and complementary information from various data representations.

Elie Karam

Scroll