Detection of Leaf Blight Disease in Sorghum Using Convolutional Neural Network

The condition known as leaf blight significantly affects sorghum, and if it isn’t treated right away, it can negatively impact the country’s economy and production. This disease affects Arbaminch Gamo Gofa and other humid regions of Ethiopia where sorghum is grown. Sorghum leaf blight is typically detected through physical inspection and chemical analysis. These methods, nevertheless, are ineffective, expensive, time-consuming, and require a specialist in the field. The newly developed model, which is based on the AlexNet pre-trained model, is utilized to identify the leaf blight disease in sorghum. The main goal of this study is to detect sorghum leaf blight disease, which affects the sorghum plant’s leaves. In the Gamo Gofa zone, which is where more sorghum is produced, the first 2000 original digital photos were collected. Following that, noise reduction and image reconstruction techniques were used to provide a picture for further study. Convolution, pooling, flattening, and full connection came next. The study’s models were created using Keras and Tensorflow. Our test results demonstrate that the identification of leaf blight in sorghum using a convolutional neural network model is 97% accurate overall.