A Comparative Study on Cotton Plant Disease Detection Using Transfer Learning Techniques

The conventional methods to identify plant diseases involve visual inspection which can be imprecise and result in delayed discovery of diseases. To solve these issues, the integration of Machine Learning driven systems for early disease detection offers promising results. In this work, a comparative analysis of various deep learning frameworks for disease identification in cotton plant datasets. The paper investigates the performance of three Transfer learning models such as InceptionV3, Xception, and DenseNet201 on a publicly available cotton dataset. The experimental results demonstrated that the DenseNet201 model achieves highest accuracy of 94.33 %. The developed model aids in identifying damaged cotton leaves and plants, laying the base for automated disease diagnosis across various plant species. This research provides a helpful means for improvising crop management and productivity in farming communities.