Scopus Indexed Publications

Paper Details


Title
AI-Powered Detection of Tobacco Leaf Diseases in Bangladesh for Improved Crop Management

Author
Minhazul Alam Mahin, Afrina Ashraf, Indrani Sen Toma, Lutfar Rahman Ratul, Md. Mizanur Rahman,

Email

Abstract

Tobacco is a major cash crop in Bangladesh, ranking among the top tobacco-producing countries and providing livelihood for thousands of families. But tobacco diseases like Alternaria alternata and Cercospora nicotianae, cause a serious decline in the quality and yield of production. Traditional manual methods of disease detection are labor intensive, time consuming and prone to errors. To address these challenges, this paper proposes a deep learning-based method for automatically detecting tobacco leaf diseases, focusing on Alternaria alternata, Cercospora nicotianae, and healthy leaves. A locally collected and augmented dataset was prepared and used to train five pretrained convolutional neural network (CNN) models, DenseNet201, InceptionResNetV2, Xception, VGG16, and ResNet50. Dropout rates of 0.5 and 0.2 and a batch size of 16 were applied during training to reduce overfitting. Among the models, DenseNet201 achieved the highest classification accuracy of 98.14%. Model performance was thoroughly evaluated using loss graphs, confusion matrices, precision, recall, F1-score, and overall accuracy. The results demonstrate that deep learning can reliably identify healthy and diseased tobacco leaves, offering a practical solution for early disease detection, supporting better crop management.



Keywords

Journal or Conference Name
2025 28th International Conference on Computer and Information Technology, ICCIT 2025

Publication Year
2025

Indexing
scopus