Res4net-CBAM: a deep cnn with convolution block attention module for tea leaf disease diagnosis

Early detection of tea leaf diseases is crucial for maintaining crop yield and agricultural production. However, manual inspection is a time-consuming and error-prone process, emphasizing the need for automated procedures. Deep learning methods have shown great potential in diagnosing plant leaf diseases. Convolutional Neural Networks (CNNs) outperform traditional deep learning models. However, the performance of these approaches is limited due to computational complexity, feature quality issues, and increasing feature dimensionality. In this study, we propose Res4net-CBAM, a deep convolutional neural network (CNN) specifically designed for tea leaf disease diagnosis, aiming to reduce the model's complexity and improve disease identification accuracy. The Res4net-CBAM model utilizes a residual block-based Res4net architecture with a network interactive convolutional block attention module (CBAM) to accurately extract complex features associated with different diseases. We conducted extensive experiments to compare the performance of our model with standard CNN models such as AlexNet, VGG16, ResNet50, DenseNet121, and InceptionV3, based on metrics such as accuracy, precision, recall, and F1-score. Our results demonstrate that the Res4net-CBAM model outperforms all other models, achieving an average recognition accuracy of 98.27% on self-acquired tea leaf disease data samples. Specifically, the Res4net-CBAM model achieved an average sensitivity of 98.39%, specificity of 98.26%, precision of 98.35%, and F1-score of 98.37%, while utilizing the Adagrad optimizer with a learning rate of 0.001. Moreover, our model surpasses some recent and existing works in this field, highlighting its effectiveness in diagnosing tea leaf diseases.

Automated summary

Early detection of tea leaf diseases is crucial for maintaining crop yield and agricultural production. This study proposes a deep convolutional neural network model specifically designed for tea leaf disease diagnosis, achieving improved accuracy. Experimental results demonstrate that the proposed model outperforms other standard CNN models and surpasses some recent works in this field.