车辆和人类数据集的鸟瞰图 （YOLO）

该数据集是用于航空图像训练和测试目标检测模型（尤其基于 YOLO 的模型）的高质量资源，由空中交通图像、VisDrone、用于车辆检测的环形交叉路口航空图像和 VSAI_Dataset 四个著名数据集汇集而成，专为鸟瞰角度检测车辆和人类定制；其拥有超 35,000 张图像及近 120 万个带注释实例，注释被标准化为 “小型车辆”“大型车辆”“人类” 三个简化类别；所有数据经预处理，删除重复或损坏注释，合并规范原始标签，按 70%、15%、15% 分为训练集、验证集、测试集，根据 CC BY-NC-SA 4.0 许可分发

This high-quality dataset is a resource for training and testing object detection models, particularly YOLO-based ones, using aerial imagery. It is compiled from four well-known datasets: Air Traffic Imagery, VisDrone, Aerial Imagery for Vehicle Detection at Roundabouts, and VSAI_Dataset, and is customized specifically for vehicle and human detection from a bird's-eye view perspective. The dataset contains over 35,000 images and nearly 1.2 million annotated instances, with annotations standardized into three simplified categories: "small vehicle", "large vehicle", and "human". All data has undergone preprocessing, including removal of duplicate or corrupted annotations and consolidation and standardization of original labels. It is split into training, validation, and test sets at a ratio of 70%, 15%, and 15% respectively, and is distributed under the CC BY-NC-SA 4.0 license.

搜集汇总

背景与挑战

背景概述

该数据集是一个专为鸟瞰角度检测车辆和人类设计的高质量资源，由多个著名航空图像数据集汇集而成，包含超35,000张图像和近120万个注释实例，注释标准化为'小型车辆''大型车辆''人类'三个类别。数据集已预处理并划分为训练集、验证集和测试集，适用于目标检测模型（尤其YOLO模型）的训练和测试，根据CC BY-NC-SA 4.0许可分发。