Mark—release—recapture methods (MRR) can provide reliable estimates of an insect population's vital parameters; however, transect counts are becoming preferred for endangered taxa due to concerns about potential damage to their habitats that may occur due to intensive sampling and adverse effects of handling on behavior and survival. Yet transect counts, when used without supplemental data on population parameters, have inherent limitations that lessen their usefulness for rigorous population monitoring, in particular for the estimation of generation population sizes. We revise and extend a method for estimating generation population size that parametrically couples models of abundance measurements from transect counts conducted annually with models of survival data from MRR studies only performed in some years. Extensions encompass 1) semiparametric modeling of count data, 2) accounting for spatial sampling error via two alternative formulations, and 3) a fully generalized approach to adjustment for imperfect detectability of individuals. Application of the basic estimator is illustrated using two endangered insects: Euphilotes battoides allyni (Shields) (Lepidoptera, Lycaenidae) and Trimerotropis infantilis Rentz (Orthoptera, Acrididae). Adult life span of the butterfly is no more than a couple of weeks, while adults of the grasshopper may live for several months. Despite these pronounced differences, our method proved flexible enough to fit the MRR and count data from each species and generate stable estimates of their respective generation population sizes. Simulation studies were also conducted and generally revealed low bias, with close agreement between actual and estimated population sizes.