More details about the parameters in Table 3 of your paper about rfbnet HOT 2 CLOSED

@ZhengMengbin Thank you for your attention. I actually didn't calculate the parameters by hand. What I do is just use weights.numel() to get the number of parameters of each layer and sum up, just a few lines of code.
And I think you may ignore the RFB module behind conv4_3, in this ablation study we use the basic RFB module instead of RFB-s.

from rfbnet.

Thank you very much for answering my question. I will try the numel() function to calculate automatically.
Because the VOC 2007 mAp corresponding to 34.5M in Table 3 is 80.1, compared with Table 2, I think it is without RFB_s.

from rfbnet.