To address the benchmarking needs of such problem, the large-scale video dataset FIVR-200K has been constructed. It comprises 225,960 YouTube videos ...
確定! 回上一頁