Biowulf High Performance Computing at the NIH
PolyRNN++: efficient interactive annotation of segmentation datasets

Manually labeling datasets with object masks is extremely time consuming. Polygon-RNN++ produces polygonal annotations of objects interactively using humans-in-the-loop. It employs Convolutional Neural Network encoder trained with Reinforcement Learning.

References:

Documentation
Important Notes

Interactive job
Interactive jobs should be used for debugging, graphics, or applications that cannot be run as batch jobs.

Allocate an interactive session and run the program. Sample session:

ssh -Y biowulf.nih.gov
[user@biowulf]$sinteractive  --gres=gpu:k80:1 --mem=12g
[user@@cn3316 ~]$module load PolyRNNpp
Download the source code to your current folder:
[user@@cn3316 ~]$git clone git clone https://github.com/davidjesusacu/polyrnn-pp
Run a demo:
[user@@cn3316 ~]$prpp
Using GPU implementation
Singularity: Invoking an interactive shell within container...
Singularity PolyRNNpp_gpu.sqsh:~>cd polyrnn-pp 
Singularity PolyRNNpp_gpu.sqsh:~>./src/demo_inference.sh 
When the computation is completed, find the resulting images in the folder "output" that will be created.
[user@cn3316 ~]$exit
salloc.exe: Relinquishing job allocation 46116226
[user@biowulf ~]$