Big data intelligent tourism management platform design based on abnormal behavior identification

Abstract

To ensure the safe and stable operation of tourist attractions, a big data intelligent tourism management platform based on abnormal behavior identification is proposed. A scenic area abnormal behavior recognition system is constructed by combining a climbing and painting behavior recognition method based on a regional convolutional 3D network model, as well as a trajectory analysis method based on object detection and tracking. Experimental data show that compared to the Two Stream model, the accuracy and recall of the 3D network model based on regional convolution are improved by 41.18 % and 34.85 %, respectively. The average accuracy of the proposed trajectory analysis method is 93.71 %. Compared with the track analysis method based on Hidden Markov model and the method based on sparse optical flow tracking, the accuracy of the proposed method is improved by 3.43 % and 1.71 %, respectively. In the real-time multi person abnormal behavior recognition system for tourist attractions, the number of frames per second for each behavior analysis is greater than 40. The results indicate that the proposed big data smart tourism management platform and related methods have effectively achieved abnormal behavior recognition in tourist attractions, improved accuracy and recall rates, and met the requirements of real-time multi person abnormal behavior recognition.