GuangkuoHu/objects365

Objects365是一个大规模图像数据集，专注于对象检测任务。它包含图像及其对应的标注信息，如图像ID、路径、格式、图像元数据（包括文件名、尺寸、许可证等）和标注详情（如对象类别、边界框、面积、是否拥挤、是否伪造、是否反射等）。数据集分为训练集和验证集，训练集有1,742,292个示例，验证集有80,000个示例，总大小约3.15 GB。该数据集适用于计算机视觉和机器学习中的对象识别和检测研究。

Objects365 is a large-scale image dataset focused on object detection tasks. It includes images along with corresponding annotations, such as image IDs, paths, formats, image metadata (e.g., file name, dimensions, license), and annotation details (e.g., object categories, bounding boxes, area, crowd status, fake status, reflection status). The dataset is split into training and validation sets, with 1,742,292 examples in the training set and 80,000 examples in the validation set, totaling approximately 3.15 GB in size. It is suitable for object recognition and detection research in computer vision and machine learning.