LRGB/coco_superpixels_edge_wt_only_coord_30
收藏资源简介:
数据集`coco_superpixels_edge_wt_only_coord_30`属于图机器学习任务类别,大小在1M到10M之间。数据集包含节点特征(x)、边索引(edge_index)、边属性(edge_attr)和标签(y)。数据集分为训练集、验证集和测试集,分别包含113287、5001和5001个样本。数据集的总下载大小为3256545字节,总数据集大小为3945248字节。数据集的应用领域是计算机视觉,任务是节点预测,节点特征包括像素和坐标(14维),边特征包括边权重(1或2维),性能评估指标是macro F1。数据集包含123,286个图,58,793,216个节点,平均每个图有476.88个节点,平均度数为5.65,总边数为332,091,902,平均每个图有2,693.67条边,平均最短路径为10.66±0.55,平均直径为27.39±2.14。数据集的创建者是Vijay Prakash Dwivedi,引用信息为Dwivedi等人于2022年发表的论文。
The dataset `coco_superpixels_edge_wt_only_coord_30` belongs to the category of graph machine learning tasks, with a size ranging from 1 MB to 10 MB. This dataset contains node features (x), edge indices (edge_index), edge attributes (edge_attr), and labels (y). It is divided into training, validation, and test sets, which contain 113287, 5001, and 5001 samples respectively. The total download size of the dataset is 3256545 bytes, and the total dataset size is 3945248 bytes. The application domain of this dataset is computer vision, and the downstream task is node prediction. The node features include pixel values and coordinates (14-dimensional), while the edge features consist of edge weights (1 or 2-dimensional). The performance evaluation metric is macro F1. This dataset contains 123,286 graphs, with a total of 58,793,216 nodes. On average, each graph has 476.88 nodes and an average degree of 5.65. The total number of edges is 332,091,902, with an average of 2,693.67 edges per graph. The average shortest path length is 10.66 ± 0.55, and the average diameter is 27.39 ± 2.14. The creator of this dataset is Vijay Prakash Dwivedi, and the citation information corresponds to the paper published by Dwivedi et al. in 2022.
数据集概述
基本信息
- 数据集名称: COCO-SP
- 任务类别: 图机器学习(graph-ml)
- 数据集大小: 1M<n<10M
- 标签: lrgb
- 许可证: CC-BY-4.0
数据集特征
- 特征名称与数据类型:
x: int64edge_index: int64edge_attr: int64y: int64
数据集划分
- 训练集:
- 样本数: 113,287
- 存储大小: 3,625,184 字节
- 验证集:
- 样本数: 5,001
- 存储大小: 160,032 字节
- 测试集:
- 样本数: 5,001
- 存储大小: 160,032 字节
数据集大小
- 下载大小: 3,256,545 字节
- 数据集总大小: 3,945,248 字节
数据集详情
- 领域: 计算机视觉
- 任务: 节点预测
- 节点特征维度: Pixel + Coord (14)
- 边特征维度: Edge Weight (1 or 2)
- 性能指标: macro F1
数据集统计
- 图数量: 123,286
- 节点总数: 58,793,216
- 平均节点数: 476.88
- 平均度数: 5.65
- 边总数: 332,091,902
- 平均边数: 2,693.67
- 平均最短路径: 10.66±0.55
- 平均直径: 27.39±2.14




