In order to solve the problem that the accuracy of human gait recognition is not high and signal features need to be extracted manually, convolutional neural network (CNN) was used to automatically extract sensor signal features to recognize five gait patterns, namely walking, going upstairs, going downstairs, going uphill and going downhill. An inertial sensor system was built to collect human motion information; according to the characteristics of the data, a fourlayer CNN model was designed to automatically extract signal features and classify gait. The feasibility of the proposed method was verified by using the measured data, and compared with the traditional recognition method of “artificial feature + support vector machine (SVM)”. The experimental results show that the proposed recognition method can be used to accurately recognize the gait, with an average recognition rate of 91.5%, and the recognition effect is better than that of the traditional scheme