Skip to content

cloudinary/dl_salicon_exercise

Repository files navigation

SALICON saliency prediction

Get the data

from the provided S3 link

Download the code

git clone https://github.com/cloudinary/dl_salicon_exercise.git

Build and run the docker

cd dl_salicon_exercise
export DATA_DIR=[parent dir of of the SALICON dataset] 
export HOST_WORKDIR=$(pwd)
docker compose -f docker-compose.yml build
docker compose -f docker-compose.yml run salicon

Run training

python train.py --config_file train_config.json --data_path /home/data/SALICON --output_path ./output

Evaluate crops

python evaluate.py --config_file train_config.json --data_path /home/data/SALICON --weight_file ./output/salicon_model.pth --output_path ./evaluate_output

Licence

Released under the MIT license.

Releases

No releases published

Packages

No packages published