Pedestrian detection is an important task in applications such as vehicle surveillance, traffic analysis, and autonomous vehicles. However, other methods for pedestrian identification in poor quality drone images pose significant challenges due to factors such as low resolution, occlusion, perspectives This paper presents a new method for identifying a are walked on the ground using the state-of-the-art YOLOv5 and YOLOv8 object detection model, which is designed for low-quality drone images Benefits of n learning and data enhancement techniques are obtained. Several tests were conducted on a standardized dataset of drone images, demonstrating the effectiveness of the method. The results show that our method outperforms the original model and gives a competitive performance, with an average accuracy (mAP) of 0.37. This work highlights the potential of applying deep learning techniques to complex real-world situations and opens the door for further research in this area.