Research

AIM Lab advances efficient, robust, and generalizable adaptation methods for vision-language models and multimodal large language models, with the goal of enabling reliable visual intelligence in real-world environments. Our research develops techniques in prompt learning, parameter-efficient adaptation, and multimodal reasoning to build AI systems that remain effective under distribution shifts. We apply these methods to high-impact domains including wildfire monitoring, advanced manufacturing, healthcare, biometrics, and digital media forensics.

Application Areas

Wildfire Monitoring

Healthcare

Media Forensics

Advanced Manufacturing

Biometrics