A knitted fabric strain sensor made from silver-plated conductive yarn and nylon spandex covered yarn is presented. Ten sensing zones are embedded into the fabricated running tights throughweft knitting technology. Physical activity tests are conducted to establish the linear relationship between the angle of the knee joint and the change in resistance of the sensing zones. A corresponding change in resistance relative knee joint movement is applied to identify three kinds of knee joint dynamics, namely, walking, sitting, and squatting, as well as four kinds of movement modes under the walking state of knee joint, including running, walking, climbing, and descending steps. The middle sensor region of the knee joint also has an obvious influence on the sensing performance of knee joint movement. Four modes of motion are extracted. Maximum resistance, average resistance, variance resistance, and median resistance are the main resistance characteristic values. Two kinds of motion cycles, namely, stand phase and swing phase, are the main time characteristic values. This thesis cites the decision tree algorithm and describes the method of realizing gait recognition.