The growing trend toward proactive health management drives the need for intelligent motion monitoring systems integrating stretchability, high sensitivity, and deep learning-driven real-time analytics for adaptive posture management. Herein, a deep learning-enabled self-powered stretchable triboelectric sensor (STS) system is introduced for intelligent posture recognition. Ti3C2Tx (MXene) nanosheets are strategically integrated into distinct layers of the STS to enhance its output performance through multifunctional mechanisms, achieving an angular resolution of ≈1° and an angle sensitivity of 0.56 V/°. By integrating the STS with an augmented reality (AR) system, real-time fitness monitoring and instant feedback are provided to facilitate continuous adjustment of users’ postures. Furthermore, an STS array integrated with a patterned shielding layer is developed to monitor and warn against fatigue states during complex cervical spine postures. Leveraging a convolutional neural network (CNN) algorithm, intelligent recognition of 14 distinct postures is achieved with 98.7% accuracy. The trained model is deployed on a smartphone to enable intelligent recognition results and reminder functions. This self-powered, deep learning-driven system bridges flexible sensing with real-time health intervention, offering scalable solutions for elderly care, sports rehabilitation, and occupational health monitoring.