Construction of intelligent gymnastics teaching model based on neural network and artificial intelligence

This study aims to develop intelligent gymnastics teaching model based on Artificial Neural Network (ANN). It addresses key issues in traditional gymnastics teaching, such as difficulty in quantifying the teaching process and lack of personalized guidance. The proposed model integrates Kinect devices and multi-sensor data. The model uses ANN to extract key features of gymnastics movements, including joint angles, velocity, and acceleration. A Hidden Markov Model (HMM) is applied to capture temporal dependencies in the movement sequences. In addition, a Graph Neural Network (GNN) is introduced to better learn complex movement patterns and improve recognition performance. The experiments used the UTKinectAction3D dataset, which contains 10 different types of gymnastics actions. The results show that the ANN-HMM-GNN model achieves excellent recognition performance. The accuracy reaches 98.2%, with a recall of 97.5% and an F1-score of 97.8%. The model also shows strong robustness and adaptability under various environmental and physical conditions. For example, under strong lighting, the accuracy, recall, and F1-score are 97.8%, 97.2%, and 97.5%, respectively. Under dim or changing lighting, performance drops slightly but remains above 95%. The model performs best with athletes of medium height (160-180 cm) and weight (60-80 kg), achieving an accuracy and F1-score of 97.5%. By combining ANN, HMM, and GNN, the proposed model not only provides accurate and robust movement recognition but also offers quantitative feedback through an intelligent feedback mechanism. This significantly improves the scientific and personalized nature of gymnastics teaching. The study offers strong technical support for the intelligent development of gymnastics education. By automatically detecting and analyzing movements, the model can deliver real-time feedback to both teachers and students, helping learners master skills more efficiently and improve performance.