ss_robust_metric2 MATLAB function performs comparison of saliency video sequence with ground-truth saliency video sequence using the method described in "EVALUATION METHODOLOGY" section of
V. Lyudvichenko, M. Erofeev, Y. Gitman and D. Vatolin, "A semiautomatic saliency model and its application to video compression," 2017 13th IEEE International Conference on Intelligent Computer Communication and Processing (ICCP), Cluj-Napoca, 2017, pp. 403-410. doi: 10.1109/ICCP.2017.8117038
The comparison carried out by the function is invariant to all common transformation, thus it allows to carry out fare comparisons of saliency models.
Note that the previous version of this algorithm does not guarantee optimal transformation and the optimization algorithm has significantly more computational complexity.
- Metric is invariant to
- Any brightness correction function
- Mixing method with Center Prior
- Finds the optimal transformation of a saliency model to fit GT saliency according to PSNR measure
The paper is available here.
More details are available at project page.
Robust Saliency Map Comparison, version 1: https://github.com/merofeev/ss_robust_metric