Addressing students by their names helps a teacher to start building rapport with students and thus facilitate their classroom participation. However, this basic yet effective skill has become rather challenging for university lecturers (especially in Asian universities), who have to handle large-sized (sometimes exceeding 100) groups in their daily teaching. To enhance lecturers' competence in delivering interpersonal interaction, we develop NaMemo, a real-time name-indicating system based on a dedicated computer vision algorithm. This paper presents its design and feasibility study, which showed a plausible acceptance level from the participating teachers and students. We also reveal students' concerns on the abuse or misuse of this system: e.g., for checking attendance. Taken together, we discuss the opportunities and risks in design, and elaborate on the plan of a follow-up, in-depth implementation to further evaluate NaMemo's impacts on learning and teaching, as well as to probe design implications including privacy considerations.