Leveraging human pose estimation for diagnostic feedback: Action research on instructional mediation and sustainable...

This study investigates the pedagogical integration of Human Pose Estimation (HPE) tools in a university-level physical fitness practicum with 31 undergraduate students, aiming to enhance students' diagnostic reasoning, reflective engagement, and instructional interaction. Grounded in a three-week action research design, the study triangulates data from HPE diagnostic reports, student reflections, and instructor journals. Results reveal that AI-assisted visual feedback, when scaffolded by deliberate instructional strategies, supported a developmental shift from surface-level tool use to conceptual understanding and applied biomechanical reasoning. Students reported increased self-efficacy and analytic precision in movement assessment, while instructors adopted adaptive pedagogical roles as mediators of learning. The integration of HPE served not only as a measurement tool but as a cognitive visualization framework that fostered reflective learning and embodied professional thinking. Theoretically, the findings extend Deep Learning Theory by demonstrating how AI-mediated visualization externalizes metacognitive processes, reframe TPACK as a dynamic and situated practice, and propose an AI-augmented Cognitive Apprenticeship model. This study contributes to sustainable pedagogy by linking AI, instructional design, and reflective practice, aligning with SDG 3 (Health), SDG 4 (Quality Education), and SDG 10 (Reduced Inequalities). It offers a replicable model for leveraging emerging AI technologies to promote inclusive, learner-centered, and future-ready teaching in physical education and beyond.