首页> 外国专利> Method, apparatus and apparatus for training neural network and detecting eye opening/closing state

Method, apparatus and apparatus for training neural network and detecting eye opening/closing state

机译:用于训练神经网络的方法,装置和装置,检测眼睛开口/关闭状态

摘要

An embodiment of the present invention discloses a training method for a neural network, a method for detecting eye opening/closing state, an intelligent driving control method, apparatus, electronic device, computer readable storage medium, and a computer program, and the training method for a neural network includes eye opening and closing of a training target. performing eye open/close state detection processing for each of a plurality of eye images in an image set corresponding to each of two or more eye open/close detection training tasks through a neural network for detection, and output a detection result of eye open/close state; Based on the eye image, the eye opening/closing labeling information and the detection result of the eye opening/closing state output from the neural network, each loss corresponding to each of the two or more eye opening/closing detection training tasks is determined, and the 2 Each of the more than one eye opening and closing detection training task is at least partially different from the eye images included in the different image sets, including the ability to adjust the network parameters of the neural network based on the corresponding losses.
机译:本发明的一个实施例公开了一种用于神经网络的训练方法,一种用于检测眼睛打开/关闭状态的方法,智能驱动控制方法,设备,电子设备,计算机可读存储介质以及计算机程序以及训练方法对于神经网络,包括训练目标的眼睛开口和关闭。在与用于检测的神经网络中的两个或多个眼睛打开/关闭检测任务中的每一个对应于对应的图像集中的每个眼睛图像中的每一个对对应的眼睛打开/关闭状态检测处理,并输出眼睛打开的检测结果/关闭状态;基于眼睛图像,眼睛打开/关闭标记信息和从神经网络输出的眼睛打开/关闭状态的检测结果,确定对应于两个或多个眼睛打开/关闭检测训练任务中的每一个的每个损失,并且,两个眼睛打开和关闭检测训练任务中的每一个至少部分地与包括在不同图像集中的眼睛图像不同,包括基于相应的损耗来调整神经网络的网络参数的能力。

著录项

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