affiliation not provided to SSRN
Traffic video segmentation \sep zero-shot learning \sep language-driven perception \sep multimodal large language models \sep training-free framework
