Within our approach, the bottom-up mechanism. (based on Faster R-CNN) proposes image regions, each with an associated feature vector, while the top-down mech-.
確定! 回上一頁