Title: Tea disease recognition technology based on a deep convolutional neural network feature learning method

Authors: Yuhan Feng

Addresses: School of Information Engineering, Xinyang Agriculture and Forestry University, Xinyang City, Henan Province, 464000, China

Abstract: China is a large tea country. Adopting more advanced science and technology to realise the intelligent identification of tea diseases will help to improve the planting, production and management of tea and carry out effective prevention and control of tea diseases. This research is based on the feature learning method of a deep convolutional neural network, introduces the optimal design of a high-order residual module, proposes an HRN algorithm, and combines self-attention mechanisms to improve the robustness of the HRN algorithm model. Through the simulation and comparative analysis with the other three algorithms, it can be seen that the HRN algorithm proposed in this study has better recognition efficiency and recognition accuracy, can effectively realise the recognition of tea diseases, and can be applied to the production and planting management of tea.

Keywords: convolutional neural network; tea; diseases; feature extraction; image recognition; high-order residual module; self-attention mechanism; tea tree disease spot.

DOI: 10.1504/IJCSM.2024.136820

International Journal of Computing Science and Mathematics, 2024 Vol.19 No.1, pp.15 - 27

Received: 29 Apr 2022
Accepted: 11 Apr 2023
Published online: 22 Feb 2024 *