Gravitational microlensing may detect dark stellar remnants - black holes or neutron stars - even if they are isolated. However, it is challenging to estimate masses of isolated dark stellar remnants using solely photometric data for microlensing events. A recent analysis of OGLE-III long-timescale microlensing events exhibiting the annual parallax effects claimed that a number of bright events were due to mass-gap objects (with masses intermediate between those of neutron stars and black holes). Here, we present a detailed description of the updated and corrected method that can be used to estimate masses of dark stellar remnants detected in microlensing events given the light curve data and the proper motion of the source. We use this updated method, in combination with new proper motions from Gaia EDR3, to revise masses of dark remnant candidates previously found in the OGLE-III data. We demonstrate that masses of mass-gap and black hole events identified in the previous work are overestimated and, hence, these objects are most likely main-sequence stars, white dwarfs, or neutron stars.