Introduction: Agriculture is one of the major backbones of the Indian economy, where rice is the most prominent staple crop across the country. However, rice production has been significantly affected due to the occurrence of various plant diseases. Deep learning and machine learning have emerged as powerful solutions for computer vision-based problems. Methods: This work identifies some of the key diseases and addresses these prominent ones using a state-of-the-art deep learning model. It proposes a novel multiclass rice leaf disease recognition model named GRG-ViT, which integrates Vision Transformer (ViT), Generative Artificial Intelligence (GenAI), and Explainable Artificial Intelligence (XAI) techniques for better outcomes. The Vision Transformer-based framework is designed to capture long-range spatial dependencies in leaf images, which enhances the model’s ability to identify the subtle disease patterns. Since the dataset portrayed considerable class imbalance, a GenAI-based synthetic data generation approach is equipped in this model to create balanced training samples, which in turn improves the model’s robustness. This model also proposes a hybrid Rectified Linear Unit (ReLU)–Gaussian Error Linear Unit (GELU)-based activation mechanism to attain effective feature representation. Results and discussion: The obtained experimental results exhibit that the proposed GRG-ViT model reaches close to an overall accuracy of 96%, which outperforms conventional approaches. The incorporation of XAI methods like Gradient-weighted Class Activation Mapping (Grad-CAM) provides both interpretability and transparency by emphasizing the regions impacting the model’s actions. This research showcases the blended power of ViT, GenAI, and XAI in producing reliable and high-performing results for rice disease detection in precision agriculture.

Introduction: In the field of precision agriculture, one of the major hurdles is the early and accurate identification of plant diseases. Farmers may face serious irreversible loss in yield if there is a delay in diagnosis by even a few days. The CNN model has helped in improving the classification of plant diseases but faces difficulty in distinguishing fine-grained symptoms, has limited generalizability, and is not easily interpretable. Though the deep CNN model performs well in classifying plant diseases, the current methods have several flaws. Most importantly, existing algorithms misclassify visually complex samples due to their limited spatial differentiation of disease-specific morphological features, including lesion borders and necrotic regions. Classification reliability is further compromised by low inter-class embedding separability for visually comparable illness phenotypes. Apart from these representational problems, training instability is still a major problem for attention-based models. Combining randomly initialized attention modules with pretrained backbone networks causes this instability, which still limits practical application. Methods: We suggest EfficientNet-CBAM-Prototype (ECP-Net), a unique end-to-end deep learning architecture, to overcome these constraints. Three complementary techniques are combined into a single framework by ECP-Net. First, parameter-efficient multi-scale feature extraction is done using an EfficientNetB0 backbone. Second, joint channel-wise and spatial feature recalibration is performed using a stabilized convolutional block attention module (CBAM). Third, a dynamic prototype memory layer uses cosine similarity-based categorization and exponential moving average (EMA) updates to maintain class-representative embedding vectors. To overcome the instability caused by randomly initialized attention weights, we further propose a two-phase training strategy wherein CBAM is frozen during phase 1 to allow prototype stabilization and then jointly fine-tuned with the learning rate in phase 2. Results: Evaluated on the PlantVillage tomato subset comprising 10 disease classes across a class-balanced split of 10,000 training, 500 validation, and 500 test samples, ECP-Net achieves 98.6% test accuracy, 98.59% F1-score, and 98.65% precision with only 4.80M parameters and 85.04 ms average inference time. These results outperformed baselines including VGG16 (97.00%), ResNet50 (81.20%), MobileNetV2 (81.20%), and CNN (70.00%). Discussion: Generalization is further validated on 35 real-field tomato leaf images captured under natural, uncontrolled conditions, confirming practical deployment potential.

Tomato leaf disease poses a significant threat to global agricultural productivity, underscoring the need for accurate and automated segmentation techniques for early detection and intervention. In this study, we proposed DSGSU-Net, an enhanced U-Net-based architecture explicitly designed for the precise segmentation of tomato leaf diseases. The model incorporates depthwise separable convolutions for efficient feature extraction, dilated convolutions in deeper layers for multi-scale context aggregation, and Ghost Sampling in the decoder for improved upsampling. To further enhance segmentation performance, a hybrid loss function combining Dice Loss and Focal Loss is utilized to manage class imbalance and enhance the boundary delineation. Experiments conducted on the PlantVillage dataset (bacterial spot class) demonstrated that DSGSU-Net achieved an accuracy of 0.9572, an F1-score of 0.8276,precision of 0.7156,recall of 0.9885, IoU of 0.7102, and a Dice coefficient of 0.9822. The results show that DSGSU-Net outperforms conventional U-Net models in segmentation accuracy and computational efficiency, making it a strong contender for practical use in precision agriculture and disease surveillance.