Resolving Ambiguity in Composed Image Retrieval via Calibrated Interaction
Researchers have introduced a new framework for Composed Image Retrieval (CIR) that addresses the inherent ambiguity in queries. Unlike previous systems that assume a single target image, this approach reframes CIR as calibrated intent resolution under uncertainty. It uses conformal prediction to provide a candidate set with a coverage guarantee, and when ambiguity is high, it asks the most informative clarifying question to narrow down the options. AI
IMPACT Introduces a novel approach to image retrieval that could improve user interaction and accuracy in complex search tasks.