HQSEG-44K
收藏魔搭社区2026-05-23 更新2024-05-15 收录
下载链接：
https://modelscope.cn/datasets/PAI/HQSEG-44K
下载链接
链接失效反馈官方服务：
资源简介：
多个对象的准确分割对于各种场景理解应用至关重要，例如图像/视频处理，机器人感知和AR/VR。SAM (Segment Anything Model)是一种用于广义图像分割的基本视觉模型。它是用十亿尺度的掩码标签训练的。通过使用点序列、边界框或粗掩码作为输入，SAM可以在多个上下文中分割各种对象、组件和视觉结构。它的零镜头分割功能已经引发了一个快速的范式变化，因为它们可以在许多应用中使用
Accurate segmentation of multiple objects is critical for a wide range of scene understanding applications, such as image/video processing, robotic perception, and AR/VR. SAM (Segment Anything Model) is a foundational visual model for generalized image segmentation. It is trained on billion-scale mask labels. By taking point sequences, bounding boxes, or coarse masks as inputs, SAM can segment various objects, components, and visual structures across diverse contexts. Its zero-shot segmentation capabilities have sparked a rapid paradigm shift, as they can be applied in numerous applications.
提供机构：
maas
创建时间：
2023-09-21
