Maize-Weed Image Dataset

The dataset contains images of maize plants and weed species. The dataset contains 36874 images in total and is stored in four folders namely Annotated Maize-Weed Images, Data Description and Questionnaire, Dry Season Maize Weed Images, and Wet Season Maize Weed Images. The Dry Season contains 18187 images captured during the dry season farm survey, the Wet Season contains 18187 images captured during the wet season farm survey and the Annotated contains 500 annotated images selected from the Dry Season survey saved in JSON, XML, and txt format. The annotation was achieved using the Labelmg suite. The raw wet and dry seasons images have been captured using a high-resolution digital camera during the weed survey, while the annotation of the annotated image was done using the Labelmg suite. A total of 18 farm locations were visited in the North Central part of Nigeria for the data acquisition as part of an ongoing research effort on maize-weed identification in farmlands.本数据集涵盖玉米植株与杂草类群的图像数据，总样本量为36874张，存储于四个文件夹中，分别为带标注玉米-杂草图像集（Annotated Maize-Weed Images）、数据说明与问卷（Data Description and Questionnaire）、旱季玉米杂草图像（Dry Season Maize Weed Images）以及雨季玉米杂草图像（Wet Season Maize Weed Images）。其中，旱季图像集包含18187张采集于旱季农田调研的图像，雨季图像集同样包含18187张采集于雨季农田调研的图像；带标注玉米-杂草图像集则包含从旱季调研图像中筛选出的500张带标注图像，标注文件以JSON、XML与TXT格式存储。本次标注工作采用Labelmg工具完成，原始的旱季与雨季图像均是在杂草调研过程中通过高分辨率数码相机拍摄获得，而500张带标注图像的标注流程同样基于该工具实现。本次数据采集共走访尼日利亚中北部地区的18个农田点位，相关工作是当前农田玉米-杂草识别研究项目的组成部分。