Are we even using scale in L2Norm?

ShuangXieIrene / ssds.pytorch

Repository for Single Shot MultiBox Detector and its variants, implemented with pytorch, python3.

MIT License

571 stars 166 forks source link

Are we even using scale in L2Norm? #24

Closed isalirezag closed 6 years ago

isalirezag commented 6 years ago

I don not understand how the scale is used in L2Norm function I appreciate it if someone can explain it for me please

foreverYoungGitHub commented 6 years ago

This is proposed by the original ssd paper. This is used because of the different scale between different layers in the VGG. In the other networks, we prefer to use the regular batch norm rather than L2 norm.