A System for Identifying Deficiency Elements in Apple Leaf Disease Based on Deep Learning
Linping Zheng,,Wanyi Li,Nanjian Li,Canming Huang,Yiqi Lin,Jinyi Cheng,Yun Kuang,Xiaoyun Guo,Zou Ling,,,,,,,,
DOI: https://doi.org/10.53964/mit.2023006
2023-12-25
Abstract:Background: Plant diseases and pests were natural disasters that seriously threaten agricultural production. Leaf diseases posed a significant threat to the overall productivity and quality of apple orchards. Currently, the diagnosis of plant diseases in apple orchards mainly relied on manual labor, which was both time-consuming and costly. It took a considerable amount of time to train an expert capable of accurately identifying apple diseases and pests, and different crop diseases vary. However, with the use of AI technology, training could be completed in just a few hours to a day, transforming computers into experts in disease and pest identification. Although China became a major apple cultivation nation, the development data from 2021 revealed that traditional advantageous production areas were undergoing dynamic adjustments due to optimization of industrial structure, leading to a decrease in cultivated land. Conversely, the demand for apples in the Chinese market continued to rise, even as the country's population structure gradually trended towards an aging population. This situation underscored the urgent need to replace the laborious and high-cost task of identifying apple foliar diseases in the cultivation process with intelligent agricultural technological solutions. Taking into account the aforementioned factors of the natural environment and socio-cultural considerations and addressing the issue of reduced apple yields due to the high impact of foliar diseases, we initiated study on disease prevention and control during the cultivation process. Methods: To overcome the objective impact of environmental factors, the system extracted images and performed operations such as enhancement and segmentation. Subsequently, a convolutional neural network was employed to process the enhanced images, extracting features such as color, shape, and texture of apple foliar lesions, which were then summarized into histograms using mathematical methods. These histograms served as the primary reference for distinguishing various diseases. Python was chosen as the programming language for the system, and the recognition system was operated by integrating frontend technologies, backend database design, and other relevant techniques. Results: Based on a substantial number of experimental results, the apple leaf disease and nutrient deficiency recognition system could accurately and with extremely high precision identifying major diseases commonly occurring in apple plants, including apple scab, leaf spot disease, rust disease, gray mold disease, and brown spot disease. Moreover, the system was capable of proposing remedies tailored to the specific pain points of each disease. It not only detected apple leaf diseases but also performs nutrient deficiency detection. This holds significant importance in enhancing both the yield and quality of apples. Conclusion: For the commonly occurring foliar diseases in cultivation, the system exhibited the capability to autonomously recognize and provide decision feedback. In contrast to the less efficient and more costly traditional approach of consulting experts, the more accessible apple foliar disease recognition system was poised to replace it as a primary tool for prevention and control assistance. This system will serve as a guiding companion, safeguarding the robust growth of apple plants.