Fetching the paper…

MAD: A Scalable Dataset for Language Grounding in Videos from Movie Audio Descriptions · Around