Filter conatenation, GoogleNet

#1

i didn’t quite understand the part u did after applying concatenation on the output of different layers in the inception module

#2

Hey @Raj.vrp, i would strongly recommed watching the video again, mentor has covered that very carefully. Also to give the idea,
supoose we have height * widht * c_channels and we want to reduce number of channels, so the mentor is suggesting to use 1*1 filter size ‘k’ filters, so that now ouput is height * width * k . This was done because “c” was very much greater, it would increase the computation cost. So we reduced it to k filter.

Hope this resolved your doubt.
Plz mark the doubt as resolved in my doubts section. :blush: