See Jerkstrand et al 2015 (Late-time spectral line formation in Type IIb SNe), section 5.5.2.
One does basically the same as for IIP SNe, with the difference that [O I] 5577 can here typically be easier measured (less blending), and knowing beta_6300 is harder observationally (because 6300 and 6364 are typically blended together) but perhaps easier theoretically (one comes much faster to optically thin emission (so beta_6300 = 1 and beta_6300/beta_5577 = 1) due to the higher expansion velocities.
From the models we found an optimal epoch to apply the method in stripped envelope SNe is ~150d. Going significantly later than that and parent state of 5577 is likely not in LTE anymore.
For the IIb SNe 1993J, 2008ax and 2011dh the O mass one obtains is 0.3-0.8 Msun, i.e. similar to IIP SNe and indicating mid-range progenitor masses MZAMS~12-18 Msun.