首页> 外国专利> CCTV CCTV System For Detecting An Action Through Real-Time Intelligent CCTV Image Analysis Using Machine Learning Object Detection And Method For Detecting An Action Through Real-Time Intelligent CCTV Image Analysis Using Machine Learning Object Detection

CCTV CCTV System For Detecting An Action Through Real-Time Intelligent CCTV Image Analysis Using Machine Learning Object Detection And Method For Detecting An Action Through Real-Time Intelligent CCTV Image Analysis Using Machine Learning Object Detection

机译：使用机器学习对象检测通过实时智能CCTV图像分析检测动作的CCTV CCTV系统以及使用机器学习对象检测通过实时CCTV图像分析检测动作的方法

页面导航

摘要
著录项
相似文献

摘要

The present invention relates to a technique for detecting a human and determining a behavior in an image captured through a CCTV. According to the present invention, a real-time intelligent CCTV image analysis behavior detection system using machine learning object detection comprises: an image capturing unit capturing a setting area and generating an image packet; an image frame generating unit receiving the image packet from the image capturing unit, and generating an image frame; an object detecting unit including a plurality of predetermined reference setting objects, receiving the image frame from the image frame generating unit, and detecting an object matched with the reference setting object in the image frame; and an object behavior classification unit including a plurality of predetermined reference behavior objects, receiving the image frame after an object is detected in the object detecting unit according to a time change, and detecting an object behavior matched with a reference behavior object in the image frame according to a time change. In addition, a real-time intelligent CCTV image analysis behavior detection method using the machine learning object detection comprises: a step (A) of enabling the image capturing unit to generate the image packet by capturing the setting area; a step (B) of enabling the image frame generating unit to generate the image frame by receiving the image packet from the image capturing unit; a step (C) of enabling the object detecting unit to detect the object matched with the reference setting object in the image frame received by the image frame generating unit, while including the plurality of predetermined reference setting objects; and a step (D) of enabling the object behavior classification unit to receive the image frame after the object is detected in the object detecting unit according to a time change, and detect the object behavior matched with the reference behavior object in the image frame according to a time change while including the plurality of predetermined reference behavior objects. The present invention detects an object and a behavior of the detected object, thereby being able to sense an accident caused by the behavior.

机译：本发明涉及一种用于检测人并确定通过CCTV捕获的图像中的行为的技术。根据本发明，一种利用机器学习对象检测的实时智能CCTV图像分析行为检测系统，包括：图像捕获单元，其捕获设置区域并生成图像包;图像帧生成单元从图像捕获单元接收图像分组，并生成图像帧;物体检测单元，其包括多个预定基准设置对象，从图像帧生成单元接收图像帧，并在图像帧中检测与基准设置对象匹配的对象;包括多个预定参考行为对象的对象行为分类单元，在对象检测单元中根据时间变化检测到对象后接收图像帧，并在图像帧中检测与参考行为对象匹配的对象行为根据时间的变化。另外，一种利用机器学习对象检测的实时智能CCTV图像分析行为检测方法，包括：步骤（A），使图像捕获单元通过捕获设置区域生成图像包;步骤（B），使图像帧生成单元通过从图像捕获单元接收图像分组来生成图像帧;步骤（C），使物体检测单元在图像帧生成单元接收到的图像帧中检测出与基准设定对象匹配的物体，同时包括多个预定基准设定对象;步骤（D），使对象行为分类单元根据时间变化在对象检测单元中检测到对象后接收图像帧，并根据图像帧检测与参考行为对象匹配的对象行为。在包括多个预定参考行为对象的同时改变时间。本发明检测物体和被检测物体的行为，从而能够感知由该行为引起的事故。

著录项

公开/公告号KR101980551B1

专利类型
公开/公告日2019-05-21

原文格式PDF
申请/专利权人 DANUSYS LTD.;
展开▼

申请/专利号KR20180136720
发明设计人 YOON IL LO;
展开▼

申请日2018-11-08
分类号G08B13/196;G06N99;H04N7/18;
国家 KR
入库时间 2022-08-21 11:48:33

相似文献

专利
外文文献
中文文献