Purpose: The purpose of this study is to explore the educational effects and assess the improvement of movements by applying an AI-based app that estimates posture and provides feedback on underhand volleyball skills to elementary school students. Methods: The research participants consisted of 177 students (age: 12 years old, height: 152.8±7.01 cm, weight: 46.72±11.2 kg, BMI: 19.90±3.68 kg/㎡) attending Elementary School located in A City. First, the volleyball underhand pass motion was analyzed to confirm the order and method for teaching the correct posture, and based on this, feedback to be used in the app was constructed. Next, through previous research on the types and operating principles of artificial intelligence posture estimation models Learned how to create an app to run on a web page. Based on this, a plan was established and an app to be used in class was developed directly. Results: The results that the experimental group showed an increase of approximately 7 minutes in practice time, whereas the comparison group showed a decrease of about 6 minutes (F = 20.873, p < .001). In terms of students' perceptions and degree of change after the app-based class, the observability domain increased after the app-based class (F = 49.216, p < .001). The testability domain also increased after the app-based class (F = 71.017, p < .001). Conclusion: Through the use of an AI posture estimation app in net-type competitive activities, students were able to perform the correct underhand volleyball technique, and their learning effects during the lessons also improved.