Running DeepHSI from a SIF File

Download the files using git and cd into the DeepHSI folder on your local computer

git clone https://github.com/shighton/DeepHSI.git && cd DeepHSI

Build a SIF container using the test.def file in DeepHSI

singularity build new.sif test.def

Copy the new SIF file into your project folder in the supercomputer (Bridges2 for this example)

scp new.sif <Username>@bridges2.psc.edu:/ocean/projects/cts090005p/<Username>

SSH into the supercomputer, navigate to your project folder, and make sure there are data and images folders

mkdir data && mkdir images

Run the new SIF file.

singularity run --bind ./data:/mnt,./:/images new.sif

The stable sif file as of June 8th 2022 can be found here or by command line:

singularity pull --arch amd64 library://andrew_satory/aum_ung_three_layer/aum-dataset:latest

Take your .sif file and place it into a new folder. Go to that folder in your terminal and run:

mkdir data

Note: if you are using the AUM dataset located in the AUM-UNG-HSI-Repository you will have to manually place the Datasets folder into the data folder you just created. Looking like /data/Datasets/AUM/gt.mat dataset.mat

Now you should run the command:

singularity run --bind ./data:/mnt test.sif

If you are using a container made using the three-layer method for instance from 5MI7th3MI6’s or semihdinc’s repository, or aum-dataset.sif downloaded above.

singularity run --bind ./data:/mnt,./:/images aum-dataset.sif

Now you can run main.py as usual for example:

python main.py --model SVM --dataset IndianPines --training_sample 0.8