AdShot: Benchmarking Multimodal Large Language Models for Video Advertisement Clipping
Advertisers routinely create multiple duration variants of ads to meet marketing budgets and viewer preferences, a labor-intensive and costly process. While multimodal large language models (MLLMs) excel at general video understanding, their ability to perform specialized editorial tasks remains unexplored. We introduc...