About the decoder part

[Jo-wang]

Hi there,

Thank you for your code.

I'm just wondering if it's a proper way to implement the decoder part using UpSampling2D since the paper mentioned that it uses max-pooling indices for upsampling.

[preddy5]

Hey @Jo-wang
I have not implemented indices-based upsampling. I have since stopped maintaining the repo :/ . You should check out https://github.com/danielenricocahall/Keras-SegNet it looks like the repo is still being maintained.