2020年使用运动摄影测量富士苹果检测和定位的注释图像和点云集合数据集

Fuji-SfM数据集可用于通过集成深度学习分割和运动结构摄影测量来检测和定位 三维 空间中的苹果。图像数据在天然果园条件下使用手持彩色相机拍摄，涉及11棵“Fuji”苹果树。数据集分为三部分，包括288张jpg格式1024×1024像素的红绿蓝图像和相应的像素级水果注释，可用于评估基于二维视觉的水果检测和分割算法；通过运动结构生成的果树3D 模型，用于多视图图像；以及带有三维边界框水果注释的扫描场景三维点云，可用于对三维水果进行检测和定位进行基准测试。http://www.grap.udl.cat/en/publications/Fuji-SfM_dataset.html

The Fuji-SfM Dataset enables the detection and localization of apples in 3D space by integrating deep learning segmentation and Structure from Motion (SfM) photogrammetry. The image data was collected using a handheld color camera under natural orchard conditions, involving 11 'Fuji' apple trees. The dataset is divided into three parts: 1. 288 RGB images in JPG format with a resolution of 1024×1024 pixels and corresponding pixel-level fruit annotations, which can be used to evaluate 2D vision-based fruit detection and segmentation algorithms; 2. 3D models of apple trees generated via SfM photogrammetry from multi-view images; 3. 3D point clouds of scanned scenes with 3D bounding box fruit annotations, which can serve as a benchmark for 3D fruit detection and localization tasks. The dataset is publicly accessible at http://www.grap.udl.cat/en/publications/Fuji-SfM_dataset.html