Damo-YOLO 算法熵值估算：基于特征图方差近似计算

Damo-YOLO 算法使用特征图的方差来近似计算网络的熵值 H(Sv)，只要其服从高斯分布就可以估算出上界。实际做法如下：

The Damo-YOLO algorithm uses the variance of the feature map to approximate the entropy value H(Sv) of the network, and an upper bound can be estimated as long as it follows a Gaussian distribution. The actual approach is as follows: