Agentic Learning for Efficient and Generalisable Visual Grounding. This project aims to develop advanced artificial intelligence systems that can understand and interpret complex visual environments m

Description

Agentic Learning for Efficient and Generalisable Visual Grounding. This project aims to develop advanced artificial intelligence systems that can understand and interpret complex visual environments more effectively, efficiently, and transparently. Current artificial intelligence models for tasks often struggle to adapt to new scenarios, require vast amounts of labeled data, and lack clarity in how decisions are made. By combining visual perception with human-like reasoning, this research will create systems that actively refine their understanding of scenes, ask questions when uncertain, and explain their decisions in plain language. The outcomes will enable safer autonomous systems and more reliable healthcare diagnostics while reducing reliance on costly data annotation.. Scheme: Discovery Early Career Researcher Award. Field: 4603 - Computer Vision and Multimedia Computation. Lead: Dr Lian Xu