Flexible piezoresistive pressure sensors integrating high sensitivity, broad linear detection range, and excellent stability are crucial for wearable electronics and human–machine interfaces. However, achieving a balanced improvement in these performance metrics remains a challenge. Herein, we propose and fabricate a high-performance flexible pressure sensor based on a chitosan/hydroxylated carbon nanotubes (CTS/CNTs-OH) composite fabric. Benefiting from the inherent antibacterial properties of chitosan and the synergistic effect with hydroxylated carbon nanotubes, the composite fabric and the sensor exhibit excellent antibacterial performance, which can effectively avoid sensor performance degradation and human skin discomfort caused by microbial growth in wearable scenarios. In addition, the chitosan-based fibrous network endows the sensor with good biocompatibility, breathability, and mechanical compliance, making it suitable for long-term skin-contact wearable applications. By optimizing the electrophoretic deposition process and the fabric layering structure, the sensor demonstrates outstanding overall performance: a broad detection range of up to 100 kPa, a maximum sensitivity of 0.151 kPa^-1, and excellent linearity (R²>0.999) across a wide pressure range. Notably, the multilayer fabric architecture enables a continuous evolution of interfacial contact under external pressure, which contributes to the simultaneous achievement of high sensitivity and a wide linear sensing range. Under dynamic pressure testing, the sensor exhibits fast response (6 ms) and recovery (46 ms) times, and it is capable of detecting subtle pressures as low as 0.98 Pa. Such rapid response characteristics allow the sensor to accurately capture both static and dynamic mechanical stimuli associated with human motions. The sensor maintains highly stable electrical output after more than 8000 loading-unloading cycles at 100 kPa, confirming its remarkable mechanical durability. For practical demonstrations, relying on its excellent antibacterial and sensing properties, the sensor was successfully employed for real-time monitoring of various human motions. Combined with a deep learning model based on a Convolutional Neural Network (CNN), it achieved high-accuracy classification of eight types of human activities, with an overall recognition accuracy exceeding 99%. These results highlight the strong compatibility between the proposed sensor and data-driven intelligent recognition algorithms. This work provides an effective material and structural design strategy for developing high-performance, wearable flexible sensing systems， especially suitable for the wearable electronics field requiring biocompatibility and antibacterial properties.