首页> 外国专利> APPARATUS FOR SORTING FOREIGN MATTERS BASED ON ARTIFICIAL INTELLIGENCE PROGRAM

APPARATUS FOR SORTING FOREIGN MATTERS BASED ON ARTIFICIAL INTELLIGENCE PROGRAM

机译:基于人工智能程序对外国事务进行排序的设备

摘要

The present invention relates to an apparatus for sorting foreign matters based on an artificial intelligence program. According to one embodiment of the present invention, the apparatus for sorting foreign matters based on an artificial intelligence program comprises: a supply unit (100) configured to supply a sorted object (1) including a plurality of pellets; a transfer unit (20) configured to receive the sorted object (1) and transfer the same by a predetermined distance; a photographing unit (30) including at least one camera for photographing the transferred sorted object (1) and configured to obtain an original image (2); a transmitting unit (40) configured to covert a size of the obtained original image (2) into a predetermined size and transmit a converted image (3); a receiving unit (50) configured to receive the converted image (3); a region extracting unit (60) configured to obtain a first feature map from the received converted image (3) using a first convolutional neural network (CNN) and extract a region of interest including the pellets from the converted image (3) by applying a region proposal network (RPN) to the first feature map; a determining unit (70) configured to determine normal pellets and abnormal pellets by applying a second CNN to an extracted region of interest image (4); and an information outputting unit (80) configured to display the region of interest on the original image (2) and integrate and output information about the abnormal pellets. According to the present invention, it is possible to quickly and accurately sort, in a work site, abnormal pallets in large quantity.;COPYRIGHT KIPO 2019
机译:基于人工智能程序的异物分类装置技术领域本发明涉及一种基于人工智能程序的异物分类装置。根据本发明的一个实施例,一种用于基于人工智能程序对异物进行分类的设备包括:供给单元(100),其被配置为供给包括多个颗粒的分类对象(1);以及供给单元(100)。传送单元(20),其被配置为接收分类的物体(1)并将其传送预定距离;拍摄单元(30),其包括至少一个用于拍摄所转移的分类对象(1)的相机,并被配置为获得原始图像(2);发送单元(40),其被配置为将所获得的原始图像(2)的尺寸转换为预定尺寸,并发送转换后的图像(3);接收单元(50),被配置为接收转换后的图像(3);区域提取单元(60),被配置为使用第一卷积神经网络(CNN)从接收到的转换图像(3)中获得第一特征图,并通过应用转换图像从转换图像(3)中提取包括颗粒的目标区域。区域提议网络(RPN)到第一个特征图;确定单元(70),其被配置为通过将第二CNN应用于提取的关注区域图像(4)来确定正常颗粒和异常颗粒;信息输出单元(80),被配置为在原始图像(2)上显示关注区域,并整合并输出关于异常颗粒的信息。根据本发明,可以在工作现场快速准确地分拣大量异常托盘。; COPYRIGHT KIPO 2019

著录项

  • 公开/公告号KR102009757B1

    专利类型

  • 公开/公告日2019-08-12

    原文格式PDF

  • 申请/专利权人 SCIGENTEC CO. LTD.;

    申请/专利号KR20190032284

  • 发明设计人 PARK YONG JAE;

    申请日2019-03-21

  • 分类号G06T7;B07C5/342;B07C5/36;G06T7/11;

  • 国家 KR

  • 入库时间 2022-08-21 11:48:02

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号