 This paper proposes a novel method for automatically detecting penguins from aerial images. The authors have developed a deep learning model, YOLO PD, which uses a combination of multiple frequency features fusion modules and bottleneck aggregation layers to improve the detection performance of small penguins. Additionally, they have incorporated a transformer aggregation layer and efficient attention module to filter out background noise and capture global features. This has resulted in an increase in accuracy compared to other models such as faster RCNN and YOLOV-7. This article was authored by Jehui Wu, Wansu, Jamfang He, and others.