The R-CNN paper uses 2,000 proposed areas (rectangular boxes) from search selective. Then, these 2,000 areas are passed to a pre-trained CNN ...
確定! 回上一頁