Semantic Segmentation

[@nithyagopalgoswami8478]

Can you give a reminder to everyone that the Input is 3x227x227 just like you did in one of the previous videos? It would help a beginner like me to not get caught up in the dimensions in every layer. Thank you for your time and effort.

[@tanoysaha9243]

clearly explained... but how it is identifying red and blue pixels... while training where we are specifying those pixels belonging to which class.

[@ShashankSingh-vm3rk]

Link for the full series please.

[@azerioauditore511]

Great detailed explanation. Thank you sir

[@bhargavasavi]

Thank you ! All the architectures discussed in the end..SegNet, UNet and DenseNet follow encoder-decoder...... DenseNet uses Dense layers instead if Conv layers...But what is the difference between SegNet and UNet....?? One difference I can observe is SegNet uses MaxUnpooling to Upsample in Decoder part, while UNet is using Upsampling by adding feature maps from Encoder to Decoder...Can anyone explain? Again thank you for the lecture ...!

[@arshadgeo8829]

Skip connections or shortcut connections will be there in U-net architecture which will transfer the whole feature maps from compression path to expansion path

[@lordnaive]

Thank you sir. Very clearly explained

[@taiftahmid5429]

Great Demonstration! Thanks!