CReLU and BatchNorm

Question

yulizhou opened this issue 7 years ago · 3 comments

Hi, I'm reading the paper and curious about your implementation.

CReLU layer seems defined but not used. Instead, the code implements it again in the layer construction.

Also, the paper has a batch norm layer but it's not implemented.

What is the consideration for this implementation? Better performance?

Thanks

Answer 1 · 2018-03-10T05:44:05.000Z

Whichever way is ok about CReLU . With the bn layer, faceboxes should have better results, I just forget to add the batch norm layer, thanks

Answer 2 · 2018-07-20T02:48:53.000Z

@yulizhou @lxg2015
Hi, I fix network design in this repo, little changes made better performance, such as conv to conv_bn_relu~

Answer 3 · 2018-08-23T08:53:38.000Z

@xiongzihua
hi, I think the predict code is wrong.We can't resize the original image.
See this repo , the output is closer to the paper.