Accurate pedestrian data is crucial for research on pedestrians within transportation hubs. Due to the fact that surveillance cameras at large transportation hubs such as airports and high-speed rail stations are typically positioned at a certain angle, it is necessary for target detection and trajectory tracking algorithms to have excellent performance. With the continuous development of target detection and trajectory tracking algorithms, precise detection and tracking of pedestrian targets has become feasible. However, there is currently a lack of quantitative evaluation of the performance of different pedestrian detection and tracking algorithms in existing research. This study presents a corresponding pedestrian dataset for monitoring environments within transportation hubs, categorizing pedestrians in the dataset according to standing and sitting postures. Simultaneously, we selected 7 high-performing target detection algorithms and 7 trajectory tracking algorithms for testing, and compared the performance of different algorithms on the dataset. Through training and evaluation of different algorithms, this study achieves efficient detection and tracking of pedestrians within transportation hubs. The findings can serve as a valuable reference for future pedestrian detection and tracking endeavors and hold significant implications for the monitoring and control of pedestrians in transportation hubs.
|