This study evaluates the performance of 6 classical and recently published TS segmentation (TSS) algorithms on a new large HAR benchmark of 126 TS with up to 13 different activities, called MOSAD, recorded with 6 participants using ordinary smartphone sensors.