Mathilde Ørstavik

Hi again, I looked into this a bit further and I'm still confused. If I use turf. nearestPointOnLine this function also gives 3,4m while turf. pointToLineDistance says 4,3m. As far...

Also, won't looking at the data in QGIS in epsg:4326 be misleading when finding the nearest point, since the coordinate-system has different scaling in the x- and y-axis? I think...

@[rowanwins](https://github.com/rowanwins) have you had time to look further into this?

It could have something to do with the initializer. Initializing the network properly can be the difference between the network converging or diverging and giving results as the ones you...

Yes, the dataset that I have trained it on has only two classes, and it works well :)

I have a [SegNet implementation](https://github.com/mathildor/TF-SegNet/blob/master/AirNet/layers.py) based on this code where I have added functionality so that the indices are used for upsampling.

Thank you for the tip! I'll look into it, but I'm probably not gonna change the code yet as I'm currently working on an implementation of a new network. I'll...

Honestly, I'm not quite sure and I remember wondering about this as well when I implemented the network. Let me look into it a bit and maybe I can give...

From the [segnet](https://arxiv.org/pdf/1505.07293.pdf) paper I found this in the first paragraph of chapter 3: «No ReLU non-linearity is used in the decoder unlike the deconvolution network [41, 42]. This makes...