Équipe Images et Contenus (IC) : BOURBIA Salima

Thématiques de recherche : Salima Bourbia holds a PhD in Computer Science and Artificial Intelligence. Since 2024, she has been an ATER in Computer Science at the Laboratoire Informatique, Image et Interaction (L3i) at La Rochelle University. Her research focuses on artificial intelligence for 3D visual data analysis, particularly 3D point cloud quality assessment using deep learning-based 2D and 3D approaches. Her work also explores perceptual quality modeling, cross-modal learning, and model optimization for efficient and reliable 3D data processing.

Points forts des activités de recherche : My research activities focus on the automatic quality assessment of 3D point clouds, which are widely used in areas such as virtual reality, 3D modeling, remote sensing, and embedded perception systems. The objective of my work is to develop methods capable of estimating the quality of these data without requiring a perfect reference, while taking into account both geometric distortions and human perception. To this end, I have explored two complementary approaches: the use of multi-view 2D projections, which makes it possible to benefit from advances in computer vision models, and the direct analysis of 3D points in order to preserve geometric information. My contributions include the integration of visual saliency, natural scene statistics, multi-task learning, and knowledge distillation between 2D and 3D models. Overall, this work aims to design models that are more accurate, more efficient, and better suited to real-world applications.