In response to the increasing demand for healthcare devices, this study devised a method for improving the sensor performance of wearable devices that can be used during weight training. Piezo-resistive sensors based on electrically conductive fibers have fatal problems when measuring bio-signals. It means that the noise signal is also measured by the surrounding environment, such as the friction of the fabric, the change of the sensor position, and the wind. Since changes in bio-signals are minute, reliability and accuracy of sensors are greatly affected by these noise signals. Therefore, it is necessary to improve discrimination through appropriate signal processing. First, the time-domain signal was converted into a frequency-domain signal through Fast Fourier Transform(FFT) and filtered through a low-pass filter(LPF). The performance of the sensor was confirmed by comparing the actual number of exercises with the number of exercises according to the raw data and filtered data. It is expected that the piezo-resistive sensor fabrication method and signal processing method used in this study will be widely applied to help improve the discrimination and accuracy of textile-based sensors for healthcare.