A pedestrian trajectory prediction method based on improved LSTM network

Fugang Liu,Songnan Duan,Wang Juan
DOI: https://doi.org/10.1049/ipr2.12954
IF: 2.3
2023-10-13
IET Image Processing
Abstract:This work proposed a deep learning‐based method that incorporated YOLOv7, StrongSORT, and improved long short term memory (LSTM) for pedestrian trajectory prediction. YOLOv7 and StrongSORT algorithm are used to solve the problems of target switch and jump in target detection and tracking. By using the dot‐product attention unit to improve the LSTM, the mutual effects of moving pedestrian orientation and distance to improve the prediction performance of overall system are fused. Pedestrian trajectory prediction based on vision is a popular task in autopilot system. As pedestrian trajectories always cross each other, pedestrian targets will frequently obscure each other, which makes the collected pedestrian trajectories produce errors. Moreover, the interaction between pedestrians will have an impact on their trajectories in crowded areas, which leads to great challenge in trajectory prediction. In order to solve the above two problems, a pedestrian trajectory prediction method is proposed. Based on the YOLOv7 algorithm, it is connected with the StrongSORT algorithm. Then adding a feature recognition (ReID) module to the model, the problem of target switching and jumping in the process of pedestrian tracking can be solved. The dot product attention mechanism is also integrated into the long short term memory (LSTM) algorithm, the improved LSTM correlate target position and distance information for the pedestrian trajectory prediction. Extensive experiments on three most challenging datasets showed that this method improves the performance and largely reduces the model size.
computer science, artificial intelligence,engineering, electrical & electronic,imaging science & photographic technology
What problem does this paper attempt to address?