MediaEval 2015 Drone Protect Task: Privacy Protection in Surveillance Systems Using False Coloring

In this paper, we share our results for privacy protection using false coloring in surveillance systems in the Drone Protect Task. The aim is obscuring sensitive regions that are privacy related without sacrificing intelligibility and pleasantness. The idea in false coloring is transforming the colors of an image using a color palette into a different set of colors in which private information is harder to recognize. The method can be applied globally to the whole frame or to a given region of interest (ROI). The privacy protected output has a pleasant look, and if desired, it can be reversed to obtain a close approximation to the original. Benchmarking evaluations on the Mini-drone dataset show promising results especially for intelligibility and pleasantness criteria.