CoSEV: A cotton disease dataset for detection and classification of severity stages and multiple disease occurrence|农业病害检测数据集|图像分类数据集

In the field of agriculture, the advancement of early treatment methods for plant leaf diseases can be greatly improved by utilizing precise and rapid automatic detection techniques. However, two common challenges arise in real-world scenarios: first, identifying different severity stages of diseases, and second, detecting multiple pathogens simultaneously affecting a single plant leaf. Unfortunately, a major hurdle in this research area is the scarcity of publicly available datasets containing images captured under varying conditions. To tackle this issue, we introduce a dataset called CoSEV for cotton diseased leaf images. The CoSEV dataset consists of 496 filtered images of cotton leaves, captured both in controlled conditions and real-field settings using a smartphone camera. After applying augmentation techniques, the total number of images reaches 1186, encompassing a diverse range of situations, including the presence of multiple stresses co-occurring on a single leaf and the progression of disease severity. The dataset has been meticulously organized into 5 classes, with 7 categories representing different levels of cotton curl severity and coexisting diseases. The dataset is uploaded both for detection and classification. To assess the effectiveness of the CoSEV dataset, we conducted experiments using various state-of-the-art detection models. These models were thoroughly analyzed to evaluate their performance in accurately identifying and classifying the different diseases and severity stages present in the dataset.