德清县耕地非粮化遥感识别数据

用于识别德清县耕地非粮化区域，精准获取位置信息，并对非粮食作物的种植情况进行分类。是分析管理和监控耕地土地利用类型变化的基础，为识别耕地转变为非粮食生产用途的情况提供高效的发现手段与精准的识别途径。系统能够进行异常数据报警，并结合历史监测数据对耕地非粮化趋势进行预判，以便管理人员精准掌握耕地的利用状况，有效保障粮食生产安全和耕地资源合理利用。利用无人机航拍技术采集的耕地非粮化遥感影像数据，通过YOLO算法实现实时目标检测。首先将神经网络应用于2024年5月遥感影像，将其划分为固定大小的网格单元，每个单元神经网络负责预测K个类别。通过预测每个区域的概率，选取所有单元格中概率最大的类别，并将其分配给相应的网格单元。根据预测概率生成加权边界框，进行非极大值抑制（NMS）运算，以消除多余的锚点。算法会识别出概率次高的边界框，并重复相同的处理过程，直到所有不同的边界框均被筛选出来。对识别准确度进行验证，0.8~1视为识别正确，算法输出识别结果正确的要素，一一对应至非粮食作物的种植情况，主要分为大棚和其他。根据耕地非粮化地块在当前识别框中的相对位置(x1,y1),(x2,y2),(x3,y3),(x4,y4)，通过四个角标位置计算相对面积area，再结合遥感影像的分辨率和比例尺计算实际面积（平方米）。对于异常数据，系统将自动获取问题点位的坐标并展示，坐标系为CGCS2000，坐标默认为东经、北纬。将影像中的特征区域及其坐标等信息上传至数字农田平台，从而实现对耕地非粮化遥感监测识别数据的获取。

This dataset is designed to identify non-grain cultivation areas on arable land in Deqing County, accurately acquire their location information, and classify the planting conditions of non-grain crops. It constitutes the foundation for analyzing, managing and monitoring changes in arable land use types, providing efficient discovery approaches and precise identification methods for detecting the conversion of arable land to non-grain production purposes.

The system can trigger alerts for abnormal data, and predict the trend of non-grain cultivation on arable land by combining historical monitoring data, enabling managers to accurately grasp the utilization status of arable land, and effectively guarantee food production safety and the rational utilization of arable land resources.

Remote sensing imagery data of non-grain cultivation on arable land is collected via unmanned aerial vehicle (UAV) aerial photography technology, and real-time object detection is implemented through the YOLO algorithm. First, the neural network is applied to the May 2024 remote sensing imagery, which is divided into fixed-size grid cells. Each grid cell is responsible for predicting K categories. By predicting the probability of each region, the category with the highest probability across all cells is selected and assigned to the corresponding grid cell. Weighted bounding boxes are generated based on the predicted probabilities, and Non-Maximum Suppression (NMS) operation is conducted to eliminate redundant anchor boxes. The algorithm identifies the bounding box with the second-highest probability and repeats the same processing until all distinct bounding boxes are screened out.

The recognition accuracy is verified, with scores ranging from 0.8 to 1 regarded as correct recognition. The algorithm outputs correctly recognized elements, which are one-to-one mapped to the planting status of non-grain crops, mainly divided into greenhouses and others. According to the relative positions (x1,y1), (x2,y2), (x3,y3), (x4,y4) of non-grain cultivation plots in the current recognition bounding box, the relative area is calculated using the four corner coordinates, and then the actual area (in square meters) is calculated in combination with the resolution and scale of the remote sensing imagery.

For abnormal data, the system will automatically acquire and display the coordinates of the problematic points. The coordinate system is CGCS2000, and the coordinates default to east longitude and north latitude. The characteristic regions in the imagery and their coordinate information are uploaded to the Digital Farmland Platform, thereby realizing the acquisition of remote sensing monitoring and recognition data for non-grain cultivation on arable land.

搜集汇总

数据集介绍

特点

德清县耕地非粮化遥感识别数据是一个由企业自行产生的数据集，包含519条记录，每季度更新一次。该数据集主要用于识别德清县耕地非粮化区域，精准获取位置信息，并对非粮食作物的种植情况进行分类，为耕地管理和监控提供基础数据支持。