Enhancing Tea Disease Identification with Lightweight MobileNetV2

Abstract

Plant diseases in tea trees can result in significant losses in both the quality and quantity of tea production. Regular monitoring can prevent the occurrence of large-scale diseases in tea plantations. However, existing methods face challenges such as a high number of parameters and low recognition accuracy, which hinder their application for monitoring tea gardens on edge devices. This paper presents a lightweight I-MobileNetV2 model for identifying diseases in tea leaves, with the goal of addressing these challenges. The proposed method embeds a coordinate attention mechanism module into the original MobileNetV2 network, enabling the model to accurately locate disease regions. Furthermore, a multi-branch parallel convolution module is employed to extract disease features across multiple scales, which improves the adaptability of the model to different disease scales. Then an automated pruning strategy is employed to compress the model and reduce computational complexity. The results indicate that algorithm proposed surpass the original MobileNetV2 by 1.91 percentage points with an average accuracy of 96.12% based on self-built tea disease dataset, the model parameters have been reduced by 40%, making it more suitable for practical application in tea garden environments.