Semantic Event Graphs for Long-Form Video Question Answering

Aradhya Dixit, Tianxi Liang

Long-form video question answering remains challenging for modern vision-language models, which struggle to reason over hour-scale footage without exceeding practical token and com…