translation-invariant anchors(平移不变锚)
at each sliding-window location, we simultaneously predict k region proposals, so the reg layer has 4k outputs encoding the coordinates of k boxes.(在每个滑动窗口位置,我们同时预测k个区域推荐,因此reg层有4k个输出,编码k个框的坐标)
the cls layer outputs 2k scores that estimate probability of object / not-object for each proposal.(cls层输出2k个分数,估计每个推荐的目标/非目标概率)
the k proposals are parameterized relative to k reference boxes, called anchors.(k个推荐相对于k个参考框(称为锚)进行参数化)
each anchor is centered at the sliding window in question, and is associated with a sca;e and aspect ratio.(每个锚定位于所讨论的滑动窗口的中心,并与比例和纵横比相关联)
we use 3 scales and 3 aspect ratios, yielding k=9 anchors at each sliding position.(我们使用3个尺度和3个纵横比,在每个滑动位置产生k=9个锚)
for a conv feature map of a size WXH (typically ~2,400), there are WXHXk anchors in total.(对于尺寸为WXH(通常为2400)的conv特征图,总共有WXHXk个锚)
an important property of our approach is that it is translation-invarient both in terms of the anchors and the functions that compute proposals relative to the anchors.(我们方法的一个重要特性是,它在锚和计算相对于锚的建议的函数方面都是平移不变性的)