As a branch of object detection task, the detection of queuing people in the aerial images emerges as a new research direction in recent years. However, it remains challenging due to the complex background, variations in the viewpoints and densely located tiny objects. In this paper, we propose a tiny object detector called QPDet to solve the above issues. Firstly, a large-scale dataset of queuing people with dense and diverse annotations has been constructed to enrich the knowledge of this field, in which a large number of queuing people have been sorted and collected using drones. Secondly, a modified YOLOX has been utilized as our detector, which effectively detects the tiny objects in the aerial images. Finally, an adaptive soft label assignment strategy has been proposed to improve the robustness of our detector. Extensive experiments have been conducted to demonstrate the effectiveness of our approach, the results exhibit the superiority of our detector.