arxivcs.CV2026-07-17
Knowing the Self, Understanding the World: A Dual-Cognition Benchmark for UAV Spatio-temporal Reasoning with MLLMs
Like Liu, Zhengzheng Xu, Haitao He, Hongzhe Li, Shuchang Zhang, Dian Shao
Multimodal large language models have achieved strong performance across diverse vision-language tasks, yet their capabilities in UAV scenarios remain insufficiently explored. Recent UAV-oriented benchmarks have begun to evaluate MLLMs in aerial scenarios, but they typically focu…