depthwise separable convolution | Complete tensor operations for MobileNet architecture

[@willyaldaircruzbejar7324]

This is the best explanation I've found. Everyone else tends to confuse me when they talk about the operations and dimensions, but here everything is clear. Thank you.

[@thesoul2086]

Best explanation on the internet.

[@saleemabedin6206]

Very good explanation. It made the concept clear.
Thank You 💙❤

[@hasrat17]

Wooo ... Beautifully explained. Thanks

[@huzaifashah2390]

Watching it too.
I have not found such simple explanation

[@aadi7448]

Awesome video! Thank you for simplifying things so well.

[@asfakhan112]

to the point video.👏

[@deafxgod6030]

Thank you for your video, great explanation

[@situ1984]

Very awesome description

[@Vivek-wn6ii]

clear explanation, thank you so much

[@riorafe]

Your explanation is very clear! thankyou

[@whenmathsmeetcoding1836]

Glad it was helpful!

[@whataday3910]

I already waited for that theory video. Nice explanation!

[@mayankupadhyay9288]

Informative and helpful video. Great work sir

[@anirudhthatipelli8765]

Thanks for the clear explanation!

[@anasssofti9271]

Thanks for the video, i got the concept, but regarding the number of operations in 2:48 are you talking about FLOPs if so, we still need to take the addition operation into account, or are you referring to MACs

[@vineetkumarmishra2989]

Really great video. Thanks

[@PaulAcademy]

Nice Explanation! Kudos.

[@adityaghosh8601]

Salute to you.

[@mikhailgorokhov4197]

Good explanation!
Thanks!

[@senthilkumarponnurangam6800]

Nicely explained

[@omanshsharma6796]

thanks bhai

[@rustemc.3850]

Awesome explanation!

[@RaghavaIndra]

Super explanation! Keep up the good work.

[@vikztube7836]

mate where u plucking 256 from mate, thin air?

[@kondurusrikanth9657]

nice video dude

[@whenmathsmeetcoding1836]

Thanks

[@CreatingUtopia]

6:14 to 6:27 is a mystery

[@tanushsshetty3274]

Thanks sir

[@AkashBiswas-r2l]

why 8*8, not 12*12 to find the total number of multiplication ?

[@content4553]

Any suggestion on how can we deal with 1*27 kernel size for convolution on a single image?

[@aashishchaubeyschannel2676]

So many people commenting that was helpful, I have a pen and a paper with me and not able to make sense working it out. Can you please clarify the calculations? For starters to understand the traditional calculations for the convolution, please watch Andrew Ng's th-cam.com/video/KTB_OFoAQcc/w-d-xo.html

[@ganeshvernekar2797]

is kernel is similar to the tensor?