Enhancing Human Motion Recognition Through Multi-Sensor Data Fusion and Deep Learning for Smart Decision Support Systems

Human motion recognition with high accuracy is important for many applications ranging from healthcare systems and sports analysis to smart environmental setups.However, traditional methods can be sensitive to sensor noise, data variability, and real-time processing requirements.This research introduces a new multi-sensor data fusion framework integrated with deep learning to improve human movement recognition for smart decision support systems.This paper presents an innovative Bayesian Convolutional Neural Network with a Long Short-Term Memory (BCNN-LSTM) framework for temporal information with data from different sensors.Multi-level fusion including feature level and decision level proposes a contrasting approach for combining sensor data that increases robustness and generalizability.The experimental results indicate that our proposed BCNN-LSTM model provides better performance than the traditional approaches, with 8% to 10% improvements in classification accuracy, compared with the Support Vector Machine, LSTM, CNN models, and Bayesian LSTM.Future enhancement includes AI integration for enhanced motion recognition precision and generalized.