Main Objects Segmentation Dataset

The Main Objects Segmentation Dataset is designed for applications in robotics and visual entertainment, comprising a vast collection of internet-collected images with resolutions ranging from 189 x 223 to 5472 x 3648 pixels. This dataset focuses on contour and semantic segmentation of a single labeled subject in each image, providing a clear and isolated view of the primary object for detailed analysis and application.

主对象分割数据集旨在服务于机器人和视觉娱乐领域的应用，该数据集汇聚了海量由网络收集的图像，其分辨率跨度从189 x 223像素至5472 x 3648像素。数据集专注于对每张图像中单个标记主题的轮廓与语义分割，从而提供对主要对象清晰且独立的视图，便于进行深入分析与应用。