Using automated stripts: can't finish processing the algorithm.

[romina-gonzalez-musso]

Hi Team!

I’m trying to run TLSeparation on my Linux machine. I think I managed to install the package properly following some recommendations posted by @philwilkes in this issue. Actually very helpful.

But I’m not sure if the processing gets stuck or may be it just take too long to finish when using any of the automated scritps. The input I’m using for testing is an individual tree point-cloud in .asc format (70Mb). Then load it as a Numpy array for running TLSeparation.

I tried changing some of the parameters of the automated scripts, but processing takes a lot of time (>3 hours for a single tree) and it never finish. It makes me wonder if I’m using the algorithms correctly. May be I have to leave it processing more time?

I’m using a Linux Mint 21.1 machine with this general specs: AMD Ryzen 9 (24 cores), 32 GB DDR4 RAM, AMD Radeon RX 5500 GPU.

Thanks in advance! :)

[romina-gonzalez-musso]

Hi again! I think I sorted this out. I made a mistake during the installation process. I was running a TLSeparation version that ended up in this bug. So I removed the whole environment and made a new fresh install. Then checked that the file that prevent the processing going into an infinite loop was actually this one. And now everything seems to be working just fine!

So, to sum up what I did:

conda create -n tlsep2 python=2
conda activate tlsep2
cd [path to where TLSeparation 1.3.2 source file was downloaded]
pip install [name of TLSeparation 1.3.2 .zip file name]

Then installed the requirements.

pip install scipy==0.19.0
pip install pandas==0.19.2
pip install numpy==1.11.3
pip install networkx==1.11
pip install scikit-learn==0.19.1

Note that I sticked to Numpy 1.11.3. The requirements.txt specifies Numpy 1.22.0 but it gives me an error when trying to install that version in a python 2 enviroment. I think Numpy 1.22.0 only supports python 3.