A lightweight rice disease identification network based on attention mechanism and dynamic convolution

A lightweight CNN model for pepper leaf disease recognition in a human palm background

“…Various researchers have effectively utilized lightweight models for plant disease identification in publicly available datasets, yielding improved recognition outcomes. Kamal et al [ 23 ] employed a modified version of MobileNet on the PlantVillage dataset, achieving a recognition accuracy of 98.34 % for 10 plant diseases with a model size of only 13 M. Sharma et al [ 28 ] introduced the multifunctional DLMC-Net, which implements a lightweight architecture using depth-separated convolution, yielding an average accuracy of 95.49 % across different plant diseases, with a model size of 25.6 M. Thakur et al [ 36 ] developed the lightweight network VGG-ICNN, demonstrating outstanding performance on the PlantVillage dataset with a recognition accuracy of 99.16 % and a model size of 23.2 M. Yang et al [ 37 ] proposed the novel lightweight high-precision network DGLNet for rice leaf disease identification, achieving a recognition accuracy of 99.82 % on a dataset comprising 38 plant diseases, with a model size of only 13.5 M. Baser et al [ 38 ] proposed an enhanced CNN model for diagnosing ten tomato plant leaf diseases, achieving an average recognition accuracy of 98.19 % and a model size of 12.0 M. Naik et al [ 29 ] curated their own chili leaf dataset and designed a lightweight convolutional neural network model (SECNN) based on squeezing and excitation, achieving an accuracy of 98.63 % on the chili leaf dataset, with a model size of only 5.4 M.…”

A robust and light-weight transfer learning-based architecture for accurate detection of leaf diseases across multiple plants using less amount of images

“…This version classifies 98.14% with a 10-megabyte model. In their study, Yang et al. (2023) introduced DGLNet, a rice disease diagnosis network that is both lightweight and accurate.…”

“…This version classifies 98.14% with a 10-megabyte model. In their study, Yang et al (2023) introduced DGLNet, a rice disease diagnosis network that is both lightweight and accurate. The Global Attention Module (GAM) and Dynamic Representation Module (DRM) are modules inside the DGLNet framework that have a modest level of complexity.…”