Abstract
We present a benchmark suite for visual perception. The benchmark is based on more than 250K high-resolution video frames, all annotated with ground-t......
小提示：本篇文献需要登录阅读全文，点击跳转登录
Full Text Link
