This repo contains subjective ranking results on BASED and RSBlur datasets along with a blur detection metric.
git clone https://github.com/illaitar/based.git
cd based
pip install -qr requirements.txt
Run data.py
to run metrics (components) and save results to csv files.
Then use eval.py
to calculate correlations of these components or train.py
to get
correlations on cross-validation (CV) and cross-dataset (CD) splits.
python data.py
python eval.py
python train.py