README.md

SaliencyADHD

This repository contains the code to calculate Normalized Scanpath Saliency (NSS) metric for each subject of the dataset ET-ADHD, video and saliency model. The used saliency models are ViNet, DeepGazeII (static), Finegrained (static) and Spectral (static).

NSS calculations

To calculate NSS for a video and a saliency model, from the notebook folder (see calculate_nss.py for argument options):

python calculate_nss.py -V 'Present' -S 'spectral'

To join all results of NSS for each saliency model:

python join_nss_df.py

To download saliency maps and results:

TBD

Directory

This repository is linked as follow

├── cache
├── data
│   ├── archive
│   └── ETFinalCutSampleEC07
│       └── ETFinalCutSample
├── notebooks
│   └── scripts
├── results
│   ├── FF
│   ├── TP
│   └── WK
├── videos_data
│   ├── Diary
│   ├── Fractals
│   └── Present
└── videos_sal
    ├── deepgazeii
    │   ├── Diary
    │   ├── Fractals
    │   └── Present
    ├── finegrained
    │   ├── Diary
    │   ├── Fractals
    │   └── Present
    ├── spectral
    │   ├── Diary
    │   ├── Fractals
    │   └── Present
    └── vinet
        ├── Diary
        ├── Fractals
        └── Present