Abstract: Objective In order to overcome the disadvantages of traditional rodent monitoring methods which were labor-intensive, time-consuming, and of low accuracy, this article used a new portable and efficient electronic monitoring system to monitor the rodents in the target areas in real time. Methods In 2016-2017, the convolutional neural network recognition technology based on deep-learning artificial intelligence was used in laboratory to recognize rodent species and analyze and record their activities, habits, and numbers in the target areas. The system box was developed after continuous adjustment and improvement. In the field application test, a categorical data analysis was performed. Results The system had over 95.00% accuracy of rodent recognition, recording, and analysis and about 90.00% counting accuracy in laboratory, which could monitor the rodent activities within 50 m in diameter in real time with clear images. In the field application, the catching rate was significantly higher than that using sticky boards. Conclusion This portable monitoring system/box realizes the intelligent real-time rodent monitoring based on the local area network, which changes the traditional rodent monitoring methods and provides a digital and smart solution of rodent surveillance information.

Key words: Rodent surveillance, Intelligent telemonitoring box, Real-time, Portability

摘要： 目的 为克服传统鼠情监测耗工、费时、精度低的缺点，使用便携高效电子监测系统箱实时监测分析区域内鼠情。方法 2016－2017年在实验室利用基于人工智能深度学习的卷积神经网络识别技术，自动甄别鼠类，记录分析鼠类活动、习性、数量等特征，经不断调整完善研制该系统箱。现场应用试验率的比较采用χ²检验。结果 该系统箱实验室对鼠类识别、记录分析准确度达到95.00%以上，计数准确度在90.00%左右。可实时监测φ50 m范围内的鼠活动，图像清晰。现场应用捕见率显著高于粘鼠板。结论 该监测系统箱基于局域网实现智能实时鼠情监测，使用便携，改变传统鼠类监测方法，为了解鼠情信息提供了智能化、信息化手段。

关键词: 鼠情, 智能监测系统箱, 实时, 便携式