Samuele Salti
Papers
6
Total Citations
78
H-Index
5
About
Samuele Salti is a computer vision and 3D perception researcher whose work spans over a decade of contributions to shape description, semantic understanding, and neural rendering. His research interests center on 3D local descriptors, point cloud processing, domain adaptation, and, more recently, Neural Radiance Fields (NeRF)-based scene reconstruction and relighting. Among his most recognized contributions is the introduction of the ReNe (Relighting NeRF) dataset, which addresses the challenging problem of novel view synthesis under unseen lighting conditions — a resource that has already garnered 37 citations since its 2023 release, reflecting its immediate relevance to the research community. His earlier work on GPU-SHOT demonstrated a commitment to making 3D descriptor matching computationally viable for real-time applications, while his research on compressed 3D descriptors tackled practical bandwidth constraints in mobile visual search. More recently, Salti has explored self-supervised learning for surface orientation via spherical CNNs and domain-shift challenges in 3D semantic segmentation — critical issues for robust deployment in autonomous driving and robotics. Across these diverse threads, his work consistently bridges theoretical rigor with real-world applicability, making him a valuable voice in the evolving landscape of 3D computer vision research.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3GPU-SHOT: Parallel Optimization for Real-Time 3D Local Description10 citations · 2013
- 4Learning to Orient Surfaces by Self-supervised Spherical CNNs9 citations · 2020
- 5Toward Compressed 3D Descriptors6 citations · 2012
- 6