首页|Patent Application Titled 'Object Detection Using Deep Learning For Real-Time St reaming Applications' Published Online (USPTO 20240071064)
Patent Application Titled 'Object Detection Using Deep Learning For Real-Time St reaming Applications' Published Online (USPTO 20240071064)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
Reporters obtained the following quote from the background information supplied by the inventors: “Neural networks are common tools for object detection. In man y circumstances, it is important for these object detection models to have a hig h detection accuracy with relatively low computational requirements. This way, t he object detection models may be used with the available hardware of machines t hat execute the object detection models. Additionally, the object detection mode ls may perform tasks in real-time and/or near real-time, which is important for many machines. For example, an object detection model that is executing on an ed ge device (e.g., a video monitor) or deployed in an autonomous or semi-autonomou s machine (e.g., vehicle, robot, aircraft, etc.) may need to detect objects and/ or the locations of objects in real-time or near real-time when performing one o r more downstream operations-such as determining the occupancy, activity, behavi or, or traffic in an area; monitoring or surveilling an area for unauthorized ac cess or behavior; detecting objects or persons in an area; or determining trajec tories for traversing an environment in which the detected objects are located.
NETL
NSTL
