Pedestrian Abnormal U-turn Behavior Recognition Model Based on Baidu AI
Wenjie Zhu, Rongyong Zhao, Chengxiao Dong, Hao Zhang, Cuiling Li, Yunlong Ma, Ping Jia · 2022
With the economic development and social progress, in order to meet people's increasing travel or entertainment needs, there are more and more crowd gathering places, such as traffic stations, venues, shopping malls and stadiums. Crowd gathering places contain many cross channels, where the anisotropy of crowd behavior is obvious, which is an important part of crowd safety monitoring. In this paper, the abnormal behavior model of the crowd in the intersection of public places is established, the abnormal U-turn behavior of pedestrians in this scene is analyzed, and the human centroid model is proposed by calling the relevant Baidu AI human key point recognition module interface. Pedestrian U-turn behavior in video is recognized by continuous frame images, and a pedestrian U-turn behavior recognition model based on Baidu AI is constructed. This study can provide targeted protection for high-risk areas in large crowd gathering places, and inform security personnel in time after identifying the abnormal behavior of pedestrians, which is helpful for man-machine cooperation to ensure the safety of pedestrians and reduce the accident rate.