Artificial intelligence (AI) recognition technology, due to its fast identification speed, good stability, and high accuracy, has been widely applied in industries such as manufacturing and agriculture. In recent years, scholars have started applying AI recognition technology to the identification of Tea plant pests and diseases, achieving improved efficiency and labor savings.

I. Overview of the Development of AI Recognition Technology

The exploration of AI recognition technology began in the 1950s with research into biological vision. Typically, this involves using image capture devices to automatically receive target images and process and analyze them. The technology is characterized by its speed, stability, and accuracy, and it has the potential to replace human eyes for identification purposes.

Entering the 21st century, traditional machine learning methods and deep learning have been widely used in AI recognition studies of agricultural pests and diseases. Early research was based on static specimen images, with less effective results in complex field environments. However, deep learning has an advantage in handling large-scale data, enabling automatic extraction of object features and classification using classifiers. Compared to traditional machine learning, deep learning offers significant improvements in accuracy and efficiency, which can significantly enhance identification accuracy and reduce labor input.

II. Current Research Status of AI Identification of Tea Plant Pests and Diseases

1. Progress in AI Identification of Tea Plant Pests and Diseases

There are over 900 recorded types of tea plant pests and diseases in China. Traditionally, their identification relied on experts and professionals who identified pests based on morphological characteristics, disease symptoms, and occurrence times. Manual identification could not meet production needs and posed challenges for precise control. In contrast, AI identification is more accurate and requires less time and labor, making it highly promising for use in identifying tea plant pests and diseases.

With the development of AI recognition technology in agricultural pest and disease identification systems, there has been progress in the study of tea plant pests and diseases. In 2008, Qin Huagang developed a web-based intelligent management system for Tea Garden pests based on expert experience. This system included three main components: pest and disease identification, pest prediction and forecasting, and pest control decision-making. It used three identification methods: morphological, spectroscopic, and retrieval. This was a representative early study that introduced AI technology into tea garden pest and disease control. In the field of image recognition, algorithms play a crucial role in identification speed and accuracy. Wu Alin et al. constructed a knowledge base for three-dimensional spatial structures of five types of tea leafworm pests using BP, SVM, and CART algorithms, achieving classification accuracy between 80.00% and 86.67%.

In recent years, convolutional neural network (CNN) technology has been widely applied in the field of image AI recognition. Models using image saliency analysis and CNNs have achieved good identification effects for common tea garden pests, improving the recognition ability for different tea plant disease images. The rapid proliferation of mobile smart devices also provides a feasible direction for pest and disease identification.

Currently, the Chinese Academy of Agricultural Sciences Tea Research Institute and Hangzhou Ruikun Technology Co., Ltd. have jointly developed a mobile-based smart recognition system called “Tea Diseases and Pests.” This system can identify about 80 common tea garden pests and natural enemies, offering simple operation, fast identification speed, and high accuracy, providing a reliable method for diagnosing tea plant pests and diseases.

2. Problems Existing in AI Identification of Tea Plant Pests and Diseases

Despite rapid advancements in AI recognition technology and improvements in deep learning applications and algorithm optimization, many issues still exist in the study of AI identification of tea plant pests and diseases.

On one hand, most research is still at the laboratory stage and does not meet practical application requirements. The main reasons include: (1) most studies are conducted indoors where external factors can be effectively controlled, but actual field conditions are more complex, with light, weather, and other factors affecting image capture; (2) laboratory studies mainly focus on static specimens, while real-world applications involve dynamic pest identification, which poses challenges and requires improved accuracy; and (3) most collected images are from advanced stages of pest or disease occurrence, whereas early detection is crucial for effective control measures.

On the other hand, software development should prioritize lightweight, simple, convenient, and easy-to-use tools to facilitate integration with various technologies.

III. Application Prospects of AI Recognition Technology in Identifying Tea Plant Pests and Diseases

Although there are some problems with the application of AI recognition technology in identifying tea plant pests and diseases, there have already been achievements in program design and implementation. In the future, these can be further developed towards monitoring, early warning, and precise control, thus advancing digital tea garden construction.

In terms of monitoring and early warning, improvements in effective algorithms will significantly increase the accuracy of pest and disease identification and the ability to grade the severity of damage. By transitioning from single-pest and disease identification to comprehensive monitoring and early warning, the advantages of AI can be fully utilized for real-time, dynamic, and integrated monitoring and early warning of tea garden pests and diseases, continuously optimizing monitoring and early warning levels and providing reliable data for tea garden pest and disease monitoring and early warning.

In terms of precise control, through long-term and multi-point intelligent monitoring data combined with local geographic positions, databases of tea garden pests, diseases, and natural enemies can be established. When outbreaks occur, based on local geographic position, climate, natural enemies, and monitoring and early warning information, timely updates on the occurrence of tea garden pests and diseases can be provided to tea farmers, enabling them to implement precise control measures and avoid the misuse of pesticides, promoting green control in tea gardens.

The tea plant pest and disease identification system is an important part of digital tea gardens. Future smart identification systems for tea gardens will not only focus on pests and diseases but can also expand to cover aspects such as tea plant cultivation and tea garden management, integrating multiple functions into a single system to improve the level of digital management in tea gardens.

This article is excerpted from “Application and Prospects of Artificial Intelligence Recognition of Tea Plant Pests and Diseases,” published in *China Tea* Issue 6, 2025, pp. 1-6. Authors: Yang Fengshui, Wang Zhibo, Wang Weton, Zhang Xinxin, Sun Liang, Xiao Qiang.