sb-ai-lab / early_hyperacute_stroke_dataset

Early hyperacute stroke dataset

Installation

Build docker image:

./build.sh

Usage

First, run container:

./run.sh %--gpus docker option value%

Then all commands are executed in the terminal of the running container. If you run training, first run MLFlow not in container:

./run_mlflow.sh

Training:

python3 early_hyperacute_stroke_dataset/scripts/train.py settings/baseline.yaml

Inference:

python3 early_hyperacute_stroke_dataset/scripts/inference.py [-h] --model MODEL --settings SETTINGS \
--dataset_metadata DATASET_METADATA --data DATA --output OUTPUT [--device_type {cpu,cuda}] [--device_num DEVICE_NUM]

Result analysis:

python3 early_hyperacute_stroke_dataset/scripts/analysis.py [-h] --predict PREDICT --reference REFERENCE --output OUTPUT

Result visualisation:

python3 early_hyperacute_stroke_dataset/scripts/vis_predict.py [-h] --predict PREDICT --reference REFERENCE \
--output OUTPUT [--show_reference]