Figure 12: Some of regions detected as interesting by our visual attention algorithm and the NVT. The videos sequences were taken approximately between 7.5 m and 11 m of depth.