Robotics & Machine Learning Daily News2024,Issue(Nov.28) :269-272.

'Determining Visual Overlap Of Images By Using Box Embeddings' in Patent Applica tion Approval Process (USPTO 20240378852)

'使用框嵌入确定图像的视觉重叠'专利申请批准程序(USPTO 20240378852)

Robotics & Machine Learning Daily News2024,Issue(Nov.28) :269-272.

'Determining Visual Overlap Of Images By Using Box Embeddings' in Patent Applica tion Approval Process (USPTO 20240378852)

'使用框嵌入确定图像的视觉重叠'专利申请批准程序(USPTO 20240378852)

扫码查看

摘要

本专利申请未转让给公司或机构。以下引文是新闻编辑从新闻编辑提供的背景资料中获得的发明人:所描述的主题一般与计算机视觉有关,特别是与确定图像的视觉重叠。“确定图像的视觉重叠对于三维(3D)重新定位和在各种应用中重建,例如增强/虚拟现实/混合现实(例如,并行现实游戏),机器人导航(例如,自驾车辆),等等。例如,3D重新定位和重建通常需要确定两幅图像描绘Same 3D表面的程度。然而,这种测定的常规方法可能非常昂贵,因为常规方法通常需要局部特征匹配和每对相对姿态的几何验证图像。当根据库计算查询图像时,成本进一步乘以,特别是当画廊里的许多图片都无关紧要。因此,需要改进图像匹配技术。

Abstract

This patent application has not been assigned to a company or institution.The following quote was obtained by the news editors from the background informa tion supplied bythe inventors: ““The subject matter described relates generally to computer vision, and in parti cular to determiningvisual overlap of images.“Determination of visual overlap of images is useful for three-dimensional (3D) re-localization andreconstruction in various applications, such as augmented/vi rtual/mixed reality (e.g., parallel reality game),robot navigation (e.g., self- driving vehicle), and so on. For example, 3D re-localization and reconstructionoften requires a determination of the extent to which two images picturing the s ame 3D surface. However,conventional methods for such a determination can be ve ry expensive, as the conventional methodstypically require local feature matchi ng and geometric verification of relative pose between every pair ofimages. The cost is further multiplied when a query image is evaluated against a gallery, e specially whenmany images in the gallery are irrelevant. Thus, improved technol ogy for image matching is needed.”

Key words

Cyborgs/Emerging Technologies/Machine Learning/Patent Application

引用本文复制引用

出版年

2024
Robotics & Machine Learning Daily News

Robotics & Machine Learning Daily News

ISSN:
段落导航相关论文