Rodolfo Lotte, Ph.D

What you mean? Perform the segmentation itself (createSuperpixelXXX() - based only in some bands), or perform the statistics for some bands? I never used this argument [-b], but I will...

You need to do some process to, first, split out all bands and, second, stack it in a new raster. After to do this process, you call createSuperpixelXXX() with this...

Perfect @cbalint13, seems correct now! Still, we have the issue of the duplicate features:  which is more related to the vector processing. As soon as I find a solution,...

Yeap, sure! As soon I have the "ComputeGeometries" method done, I will pull-request! Thank you! :)

Oh, thank you so much! Ok, I have read about the index, but I thought it could only be used for columns! So, I just tried your recommendation out and...

@scivision I'm sorry, buddy! Could you help me to sort it out? I have tried both way: ``` range_option_1 = initial_date

Hey @jaskaran1 ... As I also don't know in this version of code, try to investigate in ./decoder/fcn.py the methods **_upscore_layer** and **upsample_initilizer**. My suggestion is to plot "deconv" variable...

I'm not so sure about the first, cause it never got saved during my trainings, but... **image_iter** = interval of iterations you would like to have training images saved **eval_iter**...

Copy and rename the file encoder/fcn8_vgg.py to fcn16_vgg.py. Open the fcn16 file and change lines: `from tensorflow_fcn import fcn8_vgg` to `from tensorflow_fcn import fcn16_vgg` `vgg_fcn = fcn8_vgg.FCN8VGG(vgg8_npy_path=vgg8_npy_path)` to `vgg_fcn =...

@bendidi, complementing your last comment. I downloaded, trained and test your fork and it [did not work](https://github.com/MarvinTeichmann/KittiSeg/issues/137). Also, the script described in your comment for evaluation, was incorrect due the...