摘 要：飞鸟是飞行物中典型的“低慢小”目标，具有低可观测性，在很多场景中它又是巨大的安全隐患。所以对飞鸟进行有效的检测和驱赶是机场、高压电站等高风险区域安保工作的重心。但飞鸟种类繁杂，其自身形态变化大且机动性极强的特点，让飞鸟检测的技术难度远高于传统的目标探测。文章对常见飞鸟目标检测技术的研究及发展进行了梳理，介绍了其中各项解决方案的利弊，并提出了以帧间差分法和YOLOv5 深度学习模型为基础的新检测方案。
关键词：飞鸟检测；目标检测；运动检测；深度学习；YOLOv5
中图分类号：TP391.41；TP183；V279 文献标识码：A 文章编号：2096-4706（2021）01-0092-03
Bird Detection Algorithm Based on YOLO and Inter Frame Difference Method
YANG Luye
（Shanghai University，Shanghai 200444，China）
Abstract：Bird is a typical “low，slow and small” target in flying objects，which has low observability，and it is also a huge safety hazard in many scenes. Therefore，the effective detection of bird and driving it away is the focus of security work in high-risk areas such as airports and high-pressure power stations. But the kinds of bird are complex，their characteristic in own shape changes greatly and the mobility is very strong，which makes the technology difficulty of birds detection is more higher than the traditional target detection. This paper sorts out the research and development of common bird target detection technology，introduces the advantages and disadvantages of various solutions，and proposes a new detection scheme based on the deep learning model of inter frame difference method and YOLOv5.
Keywords：bird detection；target detection；motion detection；deep learning；YOLOv5
参考文献：
[1] 王维，项洪达. 鸟击风险计算方法研究 [J]. 中国民航大学学报，2019，37（5）：21-24.
[2] 于飞，刘东华，贺飞扬. 无人机“黑飞”对电磁空间安全的挑战 [J]. 中国无线电，2018（8）：43-44.
[3] 陈唯实. 轻小型无人机监管、探测与干扰技术 [J]. 中国民用航空，2017（7）：33-34.
[4] 张群，胡健，罗迎，等. 微动目标雷达特征提取、成像与识别研究进展 [J]. 雷达学报，2018，7（5）：531-547.
[5] REDMON J，DIVVALA S，GIRSHICK R，et al. You Only Look Once：Unified，Real-Time Object Detection [C]//2016 IEEE Conference on Computer Vision and Pattern Recognition （CVPR）.Las Vegas：IEEE，2016：779-788.
作者简介：杨陆野（1993—），男，汉族，上海人，初级工程师，硕士，研究方向：计算机视觉。