Paper Review Calls 011 -- U-Net: Convolutional Networks for Biomedical Image Segmentation

[@martazak-stepniewska892]

Karol Żak is my dad. 😍 Pozdrawiam tato.

[@nielsgefiels]

Thanks for this, the format is amazing and helped me greatly to understand the architecture and the impact it had. I am also quite interested in the 3D variation of U-NET and 3-dimensional segmentation in general, perhaps this might be something for a future call. Thanks again!

[@sylarcheng5960]

Very helpful, thanks all for sharing, I love this very much, this sharing and discussion help me out of the small space to knowing the huge ML world

[@سودانتوك]

Karol Zak many thanks for the brilliant review ,you are the best.

[@franziskaflegel7187]

Thank you for this amazing paper review!

[@zenebemarkos1437]

Smart review. Thanks all!

[@bimsarapathiraja3811]

Best discussion for unet on internet

[@mathmo]

I liked the comment about slight asymmetries!

[@richardfinney2548]

Amazing content, thank you so much

[@vijayendrasdm]

Thanks Karol for amazing explanation. I have a question though, The W(X) has two parts to it , Left part : weight map to balance class frequency Right part : contributes towards learning borders. I am not able to get any intuitions , how right part of equation helps us identify cell boundaries? I am looking for intuition for W(X). What if I don't care about boundaries? Can I ignore right part of W(X) Btw the tutorial really helped me lot. Thanks

[@omarabul-hassan6093]

I am doing a project in python using keras dealing with skull stripping. Essentially, it takes MRI scans and automatically removes the skull tissue from the MRI scan to make it easier for those who analyze these scans. Can I use semantic segmentation using keras to detect the skull and remove it?

[@LearnWithAbubakr]

Enjoyed !

[@zenebemarkos1437]

Can you share original/ground truth 30 medical images? Just I need to test/pilot my data because I'm working on limited medical images.

[@NeRusek]

You can find them in my repo under ISBI2015: github.com/karolzak/keras-unet/tree/master/input/isbi2015

[@ramchandracheke]

@@NeRusek Hi Karol, Thanks for your valuable tutorial. This is the First time, I am training a network. I have GTX 1050 gpu, 3 GB. Everything ran smoothly, but now I am getting error, ResourceExhaustedError , I searched about it. So, My gpu is not sufficient for it. Can you please suggest something. Thanks a million ! Ram

[@ericlin5841]

very helpful

[@jerryjohn1995]

It was not fair to not allowing him to present everything as he wanted. He only lose time because of the questions and conversation.