作者位次 #3/7ACL 2026Findings· Findings of ACL
K now DR - REC : Auditing Knowledge-Conditioned Visual Grounding in Referring Expression Comprehension
Guanghao Jin · Jingpei Wu · Tianpei Guo · Yiyi Niu · Weidong Zhou · Linyi Yang · Guoyang Liu
摘要
While Multimodal Large Language Models (MLLMs) have demonstrated the capacity for multi-modal reasoning, current Referring Expression Comprehension (REC) benchmarks lag behind, predominantly relying on intra-image cues and neglecting the integration of external world knowledge, which significantl...