https://doi.org/10.1140/epjs/s11734-025-01619-7
Regular Article
A two-stage classification scheme for rice leaf diseases based on the PDSwin model for practical application scenarios
College of Science, Northeast Forestry University, Hexing Road 26, 150040, Harbin, Heilongjiang Province, People’s Republic of China
Received:
27
February
2025
Accepted:
29
March
2025
Published online:
16
April
2025
Image recognition technology has been extensively applied to the identification of rice leaf diseases; however, numerous challenges persist in current methodologies. First, these approaches frequently encounter difficulties in balancing speed and accuracy. Second, they are capable of accurately identifying rice leaf diseases only within specific contexts, which hinders their practical application for swift and precise disease detection. To achieve rapid and accurate image classification, a new model named PDSwin based on Swin Transformer Tiny (Swin-T) was proposed. This model employed the improved activation function Hard-GELU and a new downsampling module PDsamp (pooled and depthwise separable convolutional sampling). Compared with Swin-T, the speed of PDSwin was increased by 15.359% and the accuracy was improved by 3.495%. Furthermore, based on the new model PDSwin, a two-stage classification method of rice leaf diseases for practical applications was proposed. This method first classified the input image backgrounds, and then selected different models to classify the diseases of rice leaves according to the outcomes of the background classification. Experiments showed high accuracy in simple (95.652%) and complex (94.818%) backgrounds. Finally, the scheme was successfully deployed on mobile devices, leading to the development of an application named “Rice Leaf Disease Classifier”. The effectiveness of this application was validated by comparing it with applications developed using other schemes and evaluating it across five rice leaf disease datasets. The results showed that this method could complete the classification task of rice diseases for practical applications quickly and accurately, and had broad application prospects.
Supplementary Information The online version contains supplementary material available at https://doi.org/10.1140/epjs/s11734-025-01619-7.
Copyright comment Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
© The Author(s), under exclusive licence to EDP Sciences, Springer-Verlag GmbH Germany, part of Springer Nature 2025
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
