Bounding Box回归与SmoothL1 Loss

R-CNN利用Bounding Box 回归得到最后的检测框，提高了检测的精度。
Fast R-CNN 使用SmoothL1 Loss取代了Bounding Box 回归。

SmoothL1 Loss 函数的实现

def _smooth_l1_loss(x, t, in_weight, sigma):
    sigma2 = sigma ** 2
    diff = in_weight * (x - t)
    abs_diff = diff.abs()
    flag = (abs_diff.data < (1. / sigma2)).float()
    y = (flag * (sigma2 / 2.) * (diff ** 2) +
         (1 - flag) * (abs_diff - 0.5 / sigma2))
    return y.sum()