SIG4VAM: Synthetic Image Generator for Visual Attention Modeling

This is a code for generating synthetic stimuli

RUN

-Run "stimulusCode.m" to generate stimuli
-Files will be saved to "dataset" and "dataset_blocks"
-both masks and extra files will be generated in such folders

GENERAL FUNCTIONS:

stimulusCode/common_param_values -> defines the experimental setup (image characteristics, object size ...)
stimulusCode/psicometric_param_values -> defines the psychometric values (psi factor, slope, N ...)
stimulusCode/general/ -> it generates any image of the selected type

Please, if you use this code, cite these two articles:

Dataset psychophysics: https://www.sciencedirect.com/science/article/pii/S0042698918302207

Dataset saliency benchmark: https://ieeexplore.ieee.org/document/9008799

Also see The CVF publication here

@article{Berga_2019_VisRes,
title = "Psychophysical evaluation of individual low-level feature influences on visual attention",
journal = "Vision Research",
volume = "154",
pages = "60 - 79",
year = "2019",
issn = "0042-6989",
doi = "https://doi.org/10.1016/j.visres.2018.10.006",
url = "http://www.sciencedirect.com/science/article/pii/S0042698918302207",
author = "David Berga and Xosé R. Fdez-Vidal and Xavier Otazu and Víctor Leborán and Xosé M. Pardo"
}
@inproceedings{Berga2019_ICCV,
  title = {SID4VAM: A Benchmark Dataset With Synthetic Images for Visual Attention Modeling},
  url = {http://dx.doi.org/10.1109/ICCV.2019.00888},
  DOI = {10.1109/iccv.2019.00888},
  booktitle = {2019 IEEE/CVF International Conference on Computer Vision (ICCV)},
  publisher = {IEEE},
  author = {Berga,  David and Vidal,  Xose Ramon Fernandez and Otazu,  Xavier and Pardo,  Xose M.},
  year = {2019},
  month = Oct,
  pages = {8788–8797}
}

NOTE: The core of the code is based on previous M.W. Spratling's Code

Ref: M. W. Spratling (2012) Predictive coding as a model of the V1 saliency map hypothesis. Neural Networks, 26:7-28. 
DOI: 10.1016/j.neunet.2011.10.002
URL=https://nms.kcl.ac.uk/michael.spratling/Code/v1_saliency.zip

SID4VAM: Synthetic Image Dataset for Visual Attention Modeling

Note: You can use this as test set and use the generator (SIG4VAM) with the automatically generated masks to train your model at wish. We use a this benchmark for saliency evaluation.

Images and Masks: https://drive.google.com/drive/folders/1YV_VGRlmPoJnYlh1W0c1REBzjQZCXQyk?usp=sharing

Fixation Data and more: https://drive.google.com/drive/folders/11VYqWsy0AY1aO7IJYqs_7ITJW8d-n5wi?usp=sharing

Results