Project contains some solutions to kaggle medical image datasets. These solutions are provided mainly as examples. All scripts are implemented as terminal programs.
Create virtual environment and install. I expect 3.11.
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
Download data from brain tumor and extract.
Run script as module
python -m scripts.brain_tumor -lr 1e-4 \
-bs 4 -e 1000 -d cuda:0 -p 30 \
-m resnet34 --dataset-root assets/Brain\ Tumor \
--outfolder runs/brain_tumor
This should produce a test accuracy of 99.20% and a test confusion matrix similar to
with training graphs
View script options with
python -m scripts.brain_tumor --help
This work is licenced with a MIT licence