首页> 外文会议>Conference on Pattern Recognition and Tracking;Society of Photo-Optical Instrumentation Engineers >Optimized training of deep neural network for image analysis using synthetic targets and augmented reality
【24h】

Optimized training of deep neural network for image analysis using synthetic targets and augmented reality

机译:使用合成目标和增强现实对图像分析进行深度神经网络的优化训练

获取原文

摘要

Acquiring large amounts of data for training and testing Deep Learning (DL) models is time consumingand costly. The development of a process to generate synthetic objects and scenes using 3D graphicssoftware is presented. By programming the path and environment in a 3D graphical engine, complexobjects and scenes can be generated for the purpose of training and testing a Deep Neural Network (DNN)model in specific vision tasks. An automatic process has been developed to label and segment objects insynthetic images and generate their corresponding ground truth files. Performances of DNNs trained withsynthetic data have been shown to outperform DNNs trained with real data.
机译:获取大量数据以训练和测试深度学习(DL)模型非常耗时 且成本高昂开发使用3D图形生成合成对象和场景的过程 介绍了软件。通过在3D图形引擎中编程路径和环境,复杂 可以生成对象和场景,以训练和测试深度神经网络(DNN) 在特定视觉任务中进行建模。已经开发了一种自动过程来标记和分段对象 合成图像并生成其相应的地面真实文件。经过DNN训练的DNN的性能 合成数据已显示出优于实际数据训练的DNN。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

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