数字视频媒体内容的自动理解与分析  

Automatic understanding and analysis of digital video media content

作　　者：唐凯[1] TANG Kai(Huai’an Senior Vocational and Technically School,Huai’an 223005,China)

机构地区：[1]淮安市高级职业技术学校,江苏淮安223005

出　　处：《无线互联科技》2024年第19期90-92,共3页Wireless Internet Science and Technology

摘　　要：文章研究了基于L2正则化优化的深度卷积神经网络在数字视频媒体内容自动理解与分析中的应用。具体来说,文章分析了视频理解与分析问题,聚焦于DCNN的理论研究,引入了L2正则化方法来对方法进行优化,在实验部分使用YouTube-VOS数据集对方法进行了验证与比较,通过F1分数和交并比指标评估了优化方法相较于标准DCNN提升效果。实验结果表明,该方法在视频对象分割任务中取得了优异的效果,验证了L2正则化在深度学习模型优化中的有效性。The article investigates the application of deep convolutional neural networks based on L2 regularization optimization in automatic understanding and analysis of digital video media content.Specifically,the article analyzed the basic principles of DCNN and introduced L2 regularization method to optimize the method.In the experimental section,the YouTube VOS dataset was used to validate and compare the method.The F1 score and intersection to union ratio index were used to evaluate the improvement effect of the optimization method compared to standard DCNN.The experimental results show that this method has achieved excellent results in video object segmentation tasks,verifying the effectiveness of L2 regularization in deep learning model optimization.

关 键 词：数字视频 视频分析 深度卷积神经网络 正则化 

分 类 号：TP37[自动化与计算机技术—计算机系统结构]