Research Article | Open Access

Volume 2023 |Article ID 0062 | https://doi.org/10.34133/plantphenomics.0062

Knowledge Distillation Facilitates the Lightweight and Efficient Plant Diseases Detection Model

Qianding Huang,^1,4 Xingcai Wu,^1,4 Qi Wang ,^1,2 Xinyu Dong,¹ Yongbin Qin,² Xue Wu,^1,3 Yangyang Gao,³ Gefei Hao ^1,3

¹State Key Laboratory of Public Big Data, College of Computer Science and Technology, Guizhou University, Guiyang 550025, China
²Text Computing & Cognitive Intelligence Engineering Research Center of National Education Ministry, Guizhou University, Guiyang 550025, China
³National Key Laboratory of Green Pesticide, Guizhou University, Guiyang 550025, China
⁴These authors contributed equally to this work

Abstract

Plant disease diagnosis in time can inhibit the spread of the disease and prevent a large-scale drop in production, which benefits food production. Object detection-based plant disease diagnosis methods have attracted widespread attention due to their accuracy in classifying and locating diseases. However, existing methods are still limited to single crop disease diagnosis. More importantly, the existing model has a large number of parameters, which is not conducive to deploying it to agricultural mobile devices. Nonetheless, reducing the number of model parameters tends to cause a decrease in model accuracy. To solve these problems, we propose a plant disease detection method based on knowledge distillation to achieve a lightweight and efficient diagnosis of multiple diseases across multiple crops. In detail, we design 2 strategies to build 4 different lightweight models as student models: the YOLOR-Light-v1, YOLOR-Light-v2, Mobile-YOLOR-v1, and Mobile-YOLOR-v2 models, and adopt the YOLOR model as the teacher model. We develop a multistage knowledge distillation method to improve lightweight model performance, achieving 60.4% mAP@ .5 in the PlantDoc dataset with small model parameters, outperforming existing methods. Overall, the multistage knowledge distillation technique can make the model lighter while maintaining high accuracy. Not only that, the technique can be extended to other tasks, such as image classification and image segmentation, to obtain automated plant disease diagnostic models with a wider range of lightweight applicability in smart agriculture. Our code is available at https://github.com/QDH/MSKD.