The basic rough sets as introduced by Pawlak have been extended in many directions. Two such extensions are the rough sets on fuzzy approximation spaces and the rough sets by De et al. in 1999 and rough sets on intuitionistic fuzzy approximation spaces by Tripathy in 2006. There are several properties of these two types of rough sets established so far. However, in this paper we show that some of the properties established are erroneous through counter examples, establish their correct versions and illustrate their application in real life situations through an example in computer vision. © Springer Nature Singapore Pte Ltd. 2017.