Weed dataset in alfalfa experimental field in the Yellow River Diversion Irrigation Area of the agro-pasture ecotone in Ningxia

This dataset is a high-quality weed object detection dataset for alfalfa experimental fields in the Yellow River irrigation area of Ningxia, aimed at supporting weed identification and control research during alfalfa planting. This dataset was collected from the alfalfa experimental site in the Yellow River irrigation area of Ningxia, covering two key growth stages: seedling and branching stages. It covers four categories of weeds, including Digitaria sanguinalis(A), Chenopodium album(B), Amaranthus tricolor(C), and other weeds(D), and includes two growth states: single plant and multiple plant. A total of 5200 high-resolution images were collected and enhanced, and all original images were accurately annotated manually using the standardized PASCAL VOC format to ensure the accuracy and consistency of the annotation information. The dataset is organized into three subdirectories: Annotations, ImageSets, and JPEGImages. The JPEGImages folder contains all 5,200 alfalfa weed images in .jpg format; the Annotations folder contains the corresponding 5,200 annotation files in .xml format; and the ImageSets folder includes the Main subdirectory, which holds train.txt, valid.txt, and test.txt files, specifying the image filenames for the training, validation, and testing sets, respectively. The dataset has been divided into a training set and a validation set in an 8:2 ratio, with rich scene diversity and target distribution characteristics. It can be widely used for model development, training, and evaluation of weed target detection tasks in alfalfa fields.