Plants are an integral part of our ecosystem, and automation in detecting their diseases has intrigued researchers all around the world. In the proposed context, the research team illustrates a comparison analysis, to detect diseased leaf images of bell pepper and tomato plants. In this research, they have developed a custom-designed CNN architecture and a deep neural network, DenseNet121. The model was evaluated using standard parameters like precision, sensitivity, specificity, F-measure, FPR, and FNR, which guarantees the outperforming ability of a pre-trained classifier with respect to the custom CNN. The balanced accuracy (BAC) of CNN and DenseNet121 was 96.5% and 98.7%, respectively, thus outperforming all other works on this particular dataset. The train data size was 80%, and the test data size was 12% with validation as 8%.
Read the complete research at www.researchgate.net.
Ghosh, Sourodip. (2021). Detecting Diseased Leaves Using Deep Learning. 10.1007/978-981-33-4866-0_6.