bdanko/loaf_resolution_512

该数据集是一个计算机视觉数据集，包含图像及其元数据，用于对象检测和分割任务。数据集特征包括图像、图像ID、文件名、宽度、高度、相机高度（图像级别）以及对象级别的注释，如注释ID、区域、边界框（bbox）、相机高度（注释级别）、类别ID、难度标志、忽略标志、人员位置、ritbox、旋转框、分割多边形和世界位置。数据集分为训练集（29,569个示例）、验证集（4,600个示例）和测试集（8,773个示例），总大小约为4.64 GB，下载大小约为4.87 GB。代码示例展示了如何加载数据集并可视化图像中的对象边界框和分割区域，暗示数据集可能用于多类别对象识别和3D位置估计。This dataset is a computer vision dataset containing images and their metadata for object detection and segmentation tasks. The features include image, image ID, file name, width, height, camera height (image level), and object-level annotations such as annotation ID, area, bounding box (bbox), camera height (annotation level), category ID, difficult flag, ignore flag, person location, ritbox, rotated box, segmentation polygon, and world location. The dataset is split into train (29,569 examples), validation (4,600 examples), and test (8,773 examples) sets, with a total size of approximately 4.64 GB and a download size of approximately 4.87 GB. The code example demonstrates how to load the dataset and visualize object bounding boxes and segmentation regions in images, suggesting potential use for multi-category object recognition and 3D location estimation.