A Benchmark of Computational Models of Saliency to Predict Human Fixation in Videos

Abstract

In many applications of computer graphics and design, robotics and computer vision, there is always a need to predict where human looks in the scene. However this is still a challenging task that how human visual system certainly works. A number of computational models have been designed using different approaches to estimate the human visual system. Most of these models have been tested on images and performance is calculated on this basis. This is twofold thesis. The first part includes saliency based object detection and enhancement using Spectral Residual Approach Phase Fourier Transform [1] implemented on images as well as on videos. As there is no benchmark on videos, to alleviate this problem we have a created a benchmark of six models implemented on 12 videos which have been viewed by 15 observers in a free viewing task in the second part. Further a weighted theory (both manual and automatic) is designed and implemented on videos using six models which improved Area under the Receiver Operating Characteristic (AU-ROC) score. We have found that GBVS and Random Centre Surround have outperformed the other models.