Nowadays many restaurants adopt a new retailing strategy of buying online and picking up in store (BOPS) to offer diversified service efficiently. During the epidemic period of coronavirus, customers behave with infection risk aversion, leading to the alteration in consuming demands for the three ways (pure online, pure offline and BOPS). Therefore, this paper establishes two models to investigate the problem on how restaurants allocate service contributions between online and offline channels. Firstly, we build utility equations to classify consumers with heterogeneous perceptions of infection risk. Considering the trait of infection risk aversion, we can obtain consuming demands of different channels and the aggregate revenues. Then, the optimal service levels of both online and offline can be calculated, as well as the optimal profits. We find that BOPS can help restaurants increase profits by adjusting the service level of both online and offline channels even in the case of consumer risk aversion. Finally, we draw out numerical experiments to verify our findings.