Video Saliency Prediction Benchmark

Explore the best methods of video saliency prediction (VSP) algorithms

This repository provides the code for work: Video Saliency Prediction Benchmark

To install the dependencies into your conda environment with python 3.8, run:

pip install -r requirements.txt

To execute benchmark calculations run:

python bench.py --models_root <path to directory with models predictions>
                --gt_root <path to directory with Ground Truth saliency maps and fixations>
                --dont_use_domain_adaptation <specifies not to use domain adaptation>
                --num_workers <number of used threads>