One concern in the algorithm

[AlanLuSun]

As we can see, the scale network branch directly regresses alpha = 1/s-1. When using Taylor expansion as shown in Eq. 8, in fact, due to sigma_0 has changed to ssigma_0, the coverage of gaussian should change from 3sigma to 3s* sigma, where is not hold in Eq. 8.

[AlanLuSun]

Eq. 8 can be found in author's paper.

[greatlog]

It can be referred to Issue#1. To be strict, the coverage of gaussian actually keeps unchanged.