Fig. 10From: Exploiting textures for better action recognition in low-quality videosPercentage improvement of BSIF-TOP over LBP-TOP and LPQ-TOP, when combined with interest point based features on HMDB-LQ and HMDB-MQBack to article page