基于语义分析和卷积神经网络的视频审查机制研究

Research on Video Censorship Mechanism Based on Semantic Analysis and Convolutional Neural Network

笔者对基于语义分析和卷积神经网络的视频审核机制进行研究,对含违禁元素的图像数据进行语义标注,利用卷积神经网络对标注数据集进行训练,得到网络模型,再将待检测视频进行抽帧截图,结合语义分析,根据截图中的违禁对象和上下帧违禁对象关联等特征得到违禁分数,最终根据分数给出该视频违禁程度建议。希望通过本文的研究,给相关研究人员带来参考和借鉴。

The author studies the video auditing mechanism based on semantic analysis and convolutional neural network,semantically labels the image data containing prohibited elements,and uses the convolutional neural network to train the labeled data set to obtain the network model,which will be detected.The video is subjected to a frame drawing,combined with semantic analysis,and the prohibited score is obtained according to the features of the prohibited object in the screenshot and the associated object of the upper and lower frames,and finally the video prohibition degree recommendation is given according to the score.I hope that through the research of this paper,the relevant researchers will bring reference and reference.