https://doi.org/10.1140/epjs/s11734-025-01731-8
Regular Article
A dual path deep-learning network with multi-scale cross attention and pyramid vision transformer for citrus leaf disease detection
1
School of Computer Science and Engineering, Vellore Institute of Technology, Chennai, India
2
Centre for Cyber Physical Systems, Vellore Institute of Technology, Chennai, India
Received:
24
February
2025
Accepted:
28
May
2025
Published online:
11
June
2025
Plant diseases represent a significant issue in the agricultural field and the automated identification of these diseases is of utmost importance for effective plant monitoring. Conventional methods can be both costly and time-intensive. However, the emergence of deep learning has introduced novel opportunities for the automated identification of plant disease patterns. This study proposes a deep learning network incorporating two parallel tracks engineered to capture distinct features from citrus leaf images. The first track utilizes an Efficient Multiscale Cross Attention that incorporates key components, such as a Multiscale Attention Module, Efficient Channel Attention Module, and a Cross-Attention Module. This track emphasizes the extraction of local features at various scales to address the inherent complexity and heterogeneity present in the data. Simultaneously, the second track leverages a Pyramid Vision Transformer (PVT) tailored for pixel-level dense prediction tasks, focusing on global feature extraction. The fusion of these two tracks combining multi-scale, local and global features, significantly captures the contextual relationships between them and potentially enhances the feature representation of the input data. To the best of our knowledge, this novel approach offers a unique integration of efficient attention mechanisms and PVT, presenting a distinctive integration for feature extraction across different scales and levels of granularity. The proposed network achieved an accuracy of 96.67% when validated on a Citrus plant dataset.
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.