Seasonal Corn Leaf Disease Dataset: A Multi-Year Collection for Robust Analysis
Description
1. The horticulture industry places a high value on corn plants because of their economic importance and widespread consumption. Nevertheless, several diseases might substantially negatively influence their well-being and output, making cultivation difficult. For a therapy to be effective, leaf diseases must be accurately and quickly identified. High-resolution images of corn plant leaves were taken from various fields site in Gurudaspur, Natore, Rajshahi, Bangladesh, and are included in this dataset. The images are divided into five distinct classes: Healthy (1038 images), Bacterial Leaf Streak (190 images), Common_rust (129 images), Gray_leaf_spot(1497 images) and Maize Chlorotic Mottle Virus (89images). These classes represent both damaged and healthy leaves. The dataset has 2943 images in total. Comprehensive comments describing the condition's nature and severity are included with every photograph. This data is essential for accurate and trustworthy model training and validation. The information also contains metadata that records the location and surrounding circumstances at the time the photograph was taken. Understanding the environmental factors influencing the prevalence of disease and enhancing the accuracy of predictive models require this contextual information. 2. At the moment, there are a lot of potential deep learning and computer vision techniques to handle these kinds of categorization and detection problems. 3. To create deep learning techniques, an extensive corn leaf disease dataset is provided. The subject matter expert from an agricultural institute collaborated with us to construct the classifications for this dataset. 4. From several different corn farms at the Gurudaspur site in Natore, Bangladesh, a total of 2943 images depicting Healthy (1038 images), Bacterial Leaf Streak (190 images), Common_rust (129 images), Gray_leaf_spot(1497 images) and Maize Chlorotic Mottle Virus (89images) were collected. Then, using methods like flipping, width shifting, height shifting, brightening, rotating, shearing, and zooming, 7500 augmented images are made from these original photos to increase the quantity of data sets.