Soil UAV Photogrammetry Dataset for Weak-Texture and Repetitive-Pattern Scenes

This dataset provides 320 UAV RGB images of a large soil-dominated area (approximately 1.5 km x 1.7 km) with weak and repetitive textures, designed for benchmarking challenging photogrammetric tasks such as feature matching, SfM, MVS, and orthophoto generation. It includes onboard GNSS/IMU-based camera exterior orientation (without RTK/PPK), pre-calibrated camera interior parameters, and a Metashape project with 4 signalized ground control points for georeferencing and accuracy evaluation. The dataset is suitable for testing reconstruction robustness in low-texture and repetitive-pattern scenes.本数据集包含320张无人机（Unmanned Aerial Vehicle, UAV）RGB影像，拍摄区域为大面积以土壤为主的场景（约1.5 km × 1.7 km），纹理微弱且重复，旨在为特征匹配、运动恢复结构（Structure from Motion, SfM）、多视图立体匹配（Multi-View Stereo, MVS）以及正射影像生成等具有挑战性的摄影测量任务提供基准测试数据。数据集涵盖基于机载全球导航卫星系统/惯性测量单元（Global Navigation Satellite System/Inertial Measurement Unit, GNSS/IMU）获取的相机外方位元素（未使用实时动态差分/后处理动态差分（Real-Time Kinematic/Post-Processed Kinematic, RTK/PPK）校正）、预校准的相机内参数，以及包含4个标记地面控制点的Metashape项目文件，可用于地理配准与精度评估。本数据集适用于测试低纹理、重复纹理场景下的重建鲁棒性。