BTLab at CVPR 2026

BTLab research was prominently showcased at CVPR 2026, the world's leading conference on artificial intelligence and computer vision, held in Denver, USA.

Prof. Marina Gavrilova delivered a keynote on fairness in medical AI models at the TRUE-V: Trustworthy, Robust, Uncertainty-aware, and Explainable Visual Intelligence workshop.

As visual AI systems continue to advance toward multimodal reasoning, embodied interaction, and real-world deployment, ensuring that these systems are trustworthy, robust, explainable, and reliable has become increasingly important.

Dr. Marina Gavrilova shared her latest research on trustworthy and explainable artificial intelligence and emphasized the importance of developing AI systems that are not only accurate and high-performing, but also transparent, interpretable, and dependable for real-world applications.

She also co-chaired the XAI4CV (Explainable AI for Computer Vision) workshop for the third consecutive year. The workshop was well attended, attracting more than 100 participants and featuring an outstanding lineup of invited speakers, reflecting the growing importance of explainable and trustworthy AI in computer vision research.

BTLab researchers MSc student Rafsan Kabir and PhD student Md Shopon presented their research paper, "ReSoFed: Reliability-Guided Model Souping for Robust Federated Learning in Heterogeneous Classroom Environments," highlighting the laboratory's latest contributions to robust federated learning.