Abstract: Bird activity poses a severe threat to aviation safety, and real-time, accurate monitoring of birds around airports is a critical issue in aviation safety that needs to be resolved. To address problems such as low location accuracy of small bird targets in infrared monitoring at airports due to background interference and low signal-to-noise ratios, this paper proposes a lightweight detection model, BISTD-YOLO, based on YOLOv8n. Firstly, to reduce model parameters and enhance the ability to capture small target details, a lightweight multi-scale convolutional module C2f-LMSC and an edge-guided module C2f-E-LMSC are designed in the Backbone and Neck structures respectively. Secondly, the network neck and detection head structures are reconstructed to retain shallow detail information while reducing model complexity. Then, a cross-space attention enhanced feature pyramid structure CSAEFPN is built. It effectively suppresses background interference and solves the feature loss problem through the synergy of BiFPN and SimAM. Finally, the normalized Wasserstein distance loss function is introduced to replace the traditional IoU metric to optimize the location accuracy of small targets. Experiments are carried out on the self-constructed AIBD. Results show that, compared to YOLOv8n, the proposed model increases mAP@0.5 and mAP@0.5:0.95 by 5.5% and 2.2%, reduces parameters by 38.2%, and is only 4MB in size. In generalization tests, it boosts mAP@0.5 by 2.5%, 1.2%, and 0.6% on NUAA-SIRST, NUDT-SIRST, and ISTD-1k respectively.