iceberg-seals.search 1.3

The ICEBERG Seals usecase package

Quality Metrics

Software Dependencies

• torch 0.4.0
• torchvision 0.2.1
• tensorboardX 1.8
• cv2
• rasterio
• affine
• pandas
• geopandas
• numpy
• PIL

Installation

PSC Bridges

From source:

$ git clone https://github.com/iceberg-project/Seals.git
$ module load cuda
$ module load python3
$ virtualenv iceberg_seals
$ source iceberg_seals/bin/activate
[iceberg_seals] $ export PYTHONPATH= <path>/iceberg_seals/bin/activate
[iceberg_seals] $ pip install . --upgrade

From PyPi:

$ module load cuda
$ module load python3
$ virtualenv iceberg_seals
$ source iceberg_seals/bin/activate
[iceberg_seals] $ export PYTHONPATH= <path>/iceberg_seals/bin/activate
[iceberg_seals] $ pip install iceberg_seals.search

To test

[iceberg_seals] $ iceberg_seals.tiling --scale_bands=299 --input_image=<image_abspath> --output_folder=./test
[iceberg_seals] $ iceberg_seals.predicting --input_image=<image_filename> --model_architecture=UnetCntWRN --hyperparameter_set=A --training_set=test_vanilla --test_folder=./test --model_path=./ --output_folder=./test_image