Horus is a model-based evaluation tool for image CVD accessibility.
First, you need to have a python3 (>=3.7) environment. You should have below packages installed.
numpy
matplotlib
pandas
PIL
sklearn
statistics
scipy
opencv-python
Then you will need to clone this github repository to finish installation.
git clone https://github.com/volcano1998/Horus.git
The pipeline Horus.py needs 3 arguments as input to run.
-i input image path you want to evaluate
-o output folder to store the evaluation result
-m pretrained model used for prediction
You can have a test run by using the following command
cd Horus
python3 Horus.py -i sample_image/sample.jpeg -o test_run/ -m pretrained_model/finalized_model.sav
where sample_image.sample.jpeg is a image we picked from a random journal, test_run/ will be the output folder, pretrained_model/finalized_model.sav is the model we provide you to use.
If you have installed the Horus successfully, you will see 6 files under test_run folder. They are
prediction.txt final prediction file, which will tell you this picture is CVD friend or not
and 5 imtermidiate results
sample_og.pdf key colors from original image
sample_og.csv key colors in RGB format from original image
sample_cb.jpeg the sample image under a CVD filter(simulated image)
sample_cb.pdf key colors from simulated image
sample_cb.csv key colors in RGB format from simulated image
The whole evaluation procedure takes about 13 seconds per picture.