Analysing inter-observer saliency variations in task-free viewing of natural images

We analyse variations of saliency between human observers when viewing natural image content. For this purpose, we conducted an eye tracking experiment to record gaze patterns of humans when viewing natural images. The resulting gaze patterns were used to create saliency maps for each observer and image. A correlation analysis is then conducted with the aim to quantify the variability of saliency between the observers and the different images. The results are considered to be a first step towards gaining a better understanding regarding the reliability of saliency maps for natural images.