In the context of increasing urban mobility, the task of accurate passenger counting becomes critically important for transportation infrastructure planning. The aim of this study is to develop an automated system for detecting and counting people in urban transport using the YOLOv8 model and depth cameras. The objectives included collecting and annotating video data, training the model, and evaluating its effectiveness. A specialized dataset was created (4,047 images, 8,918 objects), and the model was trained to achieve an F1-score of 0.90. A series of experiments with different tracking algorithms was conducted. The results confirmed the system's high accuracy and real-time applicability. The developed solution can be used for monitoring passenger flow, optimizing routes, and improving the efficiency of urban transport management.