LAVI: Lab for Artificial Visual Intelligence
Advancing visual intelligence through multimodal learning, efficient models, and trustworthy systems.
Welcome to the official webpage of Lab for Artificial Visual Intelligence (LAVI), pronounced la vie in French. At LAVI, we develop Artificial Intelligence (AI) methods, with a strong focus on computer vision and Generative AI, to address complex real-world perception problems. Our research integrates vision, language, and structured data to build robust and data-efficient models for machine perception that are capable of operating in challenging environments.
We are particularly interested in low-resource and trustworthy learning approaches, with applications in robotics, remote sensing, and medical imaging, where both accuracy and adaptability are critical.
LAVI is led by Dr. Subhankar Roy, who is aTenure-Track Assistant Professor in the Department of Management, Information and Production Engineering (DIGIP) at the University of Bergamo, Italy
Research Themes
Multimodal Learning
Learning from vision, language, and structured data jointly.
Learning with Limited Resources
Designing models that operate under limited data, computation, and memory.
Trustworthy AI
Robust, extendable, privacy preserving, and reliable machine learning systems.
Application Areas
General Computer Vision
Classical computer vision tasks: image recognition, object detection, semantic segmentation, and image generation.
Robotics
Models that enable robots to perceive, reason, and act in real-world environments.
Remote Sensing
Multimodal learning to remote sensing for robust environmental monitoring and analysis.
Medical Imaging
Applying multimodal learning to medical imaging for robust diagnosis and clinical decision support.
News
- [Jun, 2026] Our paper "ProactiveBench: Benchmarking Proactiveness in Multimodal Large Language Models" accepted at ECCV 2026!
- [Mar, 2026] Our paper "Predict-then-Diffuse: Adaptive Response Length for Compute-Budgeted Inference in Diffusion LLMs" accepted at IJCNN 2026!
- [Feb, 2026] Our paper "Organizing Unstructured Image Collections using Natural Language" accepted at CVPR Findings 2026!
- [Jan, 2026] Our paper "Ensembling Pruned Attention Heads For Uncertainty-Aware Efficient Transformers" accepted at ICLR 2026!
- [2025] 📣 We are excited to announce the launch of Lab for Artificial Visual Intelligence (LAVI)!