2019年用于水果检测富士苹果多模式图像数据集

KFuji数据集是三模态图像的集合，整合了树上“Fuji”苹果的颜色（红绿蓝三色）、深度和范围校正IR强度数据。图像数据使用Microsoft Kinect v2 相机拍摄，深度图像分辨率为512×424像素，采用jpg格式。相机安装在移动平台上，夜间人工照明。研究者对原始数据执行几何配准，以在 红绿蓝、深度和范围矫正IR 通道之间建立像素级对应关系。生成的数据集包含967个多模态图像和水果注释。数据来自12,839 个苹果。该数据集为基于 红绿蓝三色传感器的田间机器人提供了水果检测和定位算法的新基准。http://www.grap.udl.cat/en/publications/KFuji_RGBDS_database.html

The KFuji dataset is a collection of three-modal images, integrating red-green-blue (RGB) color, depth, and range-corrected IR intensity data of 'Fuji' apples growing on trees. The image data was captured using a Microsoft Kinect v2 camera, with depth images having a resolution of 512×424 pixels and stored in JPEG format. The camera was mounted on a mobile platform, and artificial lighting was utilized at night. Researchers performed geometric registration on the raw data to establish pixel-level correspondences among the RGB, depth, and range-corrected IR channels. The resulting dataset contains 967 multimodal images along with fruit annotations, and the data is sourced from 12,839 apples. This dataset provides a new benchmark for fruit detection and localization algorithms for field robots equipped with RGB sensors. http://www.grap.udl.cat/en/publications/KFuji_RGBDS_database.html