从人脸识别到“虫脸识别”，难度多大
每年有约三分之一的全球作物产量因病虫草害而损失。如何降低危害，减少损失，保障粮食安全，是农业前沿科技关注的重点领域。
About one-third of global crop production is lost each year to pests, diseases, and weeds. How to reduce hazards, reduce losses, and ensure food security are key areas of concern for cutting-edge agricultural science and technology.
人脸识别，是基于人的脸部特征信息进行身份识别的一种生物识别技术。用摄像机或摄像头采集含有人脸的图像或视频流，并自动在图像中检测和跟踪人脸，进而对检测到的人脸进行脸部识别的一系列相关技术，通常也叫做人像识别、面部识别。现在已广泛应用到人们日常生活中的很多领域上，已被人们所熟知，但是虫脸识别技术不知道有多少人听说过，也许从事农业的工作者会有所了解。“虫脸识别”是一种基于人工智能图像识别和检测技术，让机器自动化识别照片中害虫种类和数量的病虫害测报手段。经由拍摄、上传、分析、反馈等环节，可以快速了解农田内的病虫害情况。
Facial recognition is a biometric technology that identifies individuals based on their facial features. A series of related technologies, commonly known as portrait recognition or facial recognition, that capture images or video streams containing faces using a camera, automatically detect and track faces in the image, and then perform facial recognition on the detected faces. Nowadays, it has been widely applied in many fields of people's daily lives and is well-known. However, it is unknown how many people have heard of insect face recognition technology, and perhaps agricultural workers will have some understanding. "Insect face recognition" is a disease and pest prediction method based on artificial intelligence image recognition and detection technology, which enables machines to automatically recognize the types and quantities of pests in photos. By taking photos, uploading, analyzing, and providing feedback, one can quickly understand the situation of pests and diseases in farmland.
虫脸识别技术在农业领域的应用
The application of insect face recognition technology in the field of agriculture
利用具有高清摄像头与传感器的“自拍杆”和搭载专用App的智能终端专业设备，伸到作物根系、果树树梢、叶片、茎秆等部位，进行图像采集，在田间完成图像采集后，图片将通过专用App被上传至后端的算法服务器上，服务器会基于人工智能技术对这些图像中包含的信息进行分析与综合研判，并将识别结果数据返回至移动终端。在移动终端上，用户可以实时查看当前的图像中包含有哪些害虫以及害虫的数量，也可以根据多个采样点的识别结果综合评估出当前田块中可能的虫害发生等级，辅助农业植物保护专家完成快速田间调查，并提供合适的防治建议。
Using a "selfie stick" with high-definition cameras and sensors, as well as intelligent terminal professional equipment equipped with dedicated apps, it extends to crop roots, fruit tree branches, leaves, stems and other parts for image acquisition. After completing image acquisition in the field, the images will be uploaded to the backend algorithm server through a dedicated app. The server will analyze and comprehensively judge the information contained in these images based on artificial intelligence technology, And return the recognition result data to the mobile terminal. On mobile terminals, users can view in real-time which pests are included in the current image and the number of pests. They can also comprehensively evaluate the possible level of pest occurrence in the current field based on the recognition results of multiple sampling points, assist agricultural plant protection experts in completing rapid field investigations, and provide appropriate prevention and control suggestions.
图片来源：中科院合肥物质科学研究院智能机械研究所网站
虫脸识别需要突破的难题
Difficulties to be overcome in insect face recognition
1、很多害虫的相似度极高，例如鳞翅目下就包含数十种常见田间作物害虫，外貌特征很相似，“普通人看上去都是蛾子”，有些类别之间的区别仅仅在翅膀上的一个不起眼的小斑点，专业人员也需仔细分辨，因此使用人工智能进行归纳比较困难。
2、害虫大小不一，有的害虫在照片中会小到难以进行形态分辨；此外，拍摄手法导致的逆光、阴影等会让拍摄采样质量有较大波动，进一步增加了识别难度。
虫脸识别技术在农业领域的发展前景
The development prospects of insect face recognition technology in the field of agriculture
未来将通过无人设备或者更加智能化的辅助设备来替代人工完成虫害数据的采集工作，再利用先进的人工智能技术，逐步替代人工构建、补充及维护预测模型的工作，可以实现自动化的快速迭代害虫发生预测模型，帮助农业专家们更快更准确地预测病虫害发生，快速高效的提出解决方案，有效控制虫害对农作物等植物的迫害。
参考文献：
