In this paper, we developed a resilient streamline optimization method for the baby carriage in a subway station based on a video evaluation system and the simulation approach. Yolov5 algorithm is adopted to detect the streamline of the baby carriages visually. According to the movement characteristics of the baby carriage, the detected streamlines are evaluated by the analytic hierarchy process (AHP) from resilient aspects. Finally, a digital-twin simulation model, including all the facilities and people in the subway station, is built in AnyLogic to access the current streamlines of baby carriages. A feasible optimization scheme is given based on the evaluation and simulation results. The case studies of Zhuguang Road Station of Shanghai subway Line 13 are executed. Different situations are tested and simulated. More resilient streamlines for baby carriages are designed. The results showed that this approach could effectively optimize the streamlines for baby carriages.