The profit in the bus business is declining, and service improvements such as route planning and optimization are required. Information about the attributes of bus passengers is necessary to improve operational management and develop more services. In this research, a method for attribute estimation utilizing multiple images of the same passenger is proposed. Passenger attributes such as age group and gender are inferred by the Swin-Transformer-based algorithm. To evaluate the performance of the proposed approach, a bus passenger dataset is collected from cameras installed at bus entrances and exits. Experimental results on the collected dataset indicate that our proposed algorithm achieves high accuracy in most attribute categories and proves its effectiveness.