“…The objective of this study is to evaluate and compare different motion detection methods and identify the best method for different situations using a challenging complete dataset. To this end, we tested the following methods: temporal differencing (frame difference) 86,149,150 , threeframe difference (3FD) [151][152][153] , adaptive background (average filter) 90,154,155 , forgetting morphological temporal gradient (FMTG) 156 , Background estimation 157,158 , spatio-temporal Markov field [159][160][161] , running Gaussian average (RGA) 14,162,163 , mixture of Gaussians (MoG) 15,59,164 , spatio-temporal entropy image (STEI) 165,166 , difference-based spatio-temporal entropy image (DSTEI) 166,167 , eigen-background (Eig-Bg) 42,168,169 and simplified self-organized map (Simp-SOBS) 24 methods. Many of these methods (3FD, ,FMTG, STEI, DSTEI, Simp-SOBS) have not been previously evaluated on challenging dataset, to this end, we used the CDnet2012 The remainder of this paper is organized as follows.…”