基于表情识别的文娱节目效果评价方法研究  

Research on the effect evaluation method of entertainment programs based on expression recognition

作　　者：滕伟楠 TENG Weinan(School of Computer Science,Xi′an Shiyou University,Xi′an 710065,China)

机构地区：[1]西安石油大学计算机学院,西安710065

出　　处：《智能计算机与应用》2024年第8期143-150,共8页Intelligent Computer and Applications

摘　　要：本研究旨在探讨一种基于表情识别的文娱节目效果评价方法。首先,针对Emotion-Domestic数据集中丰富的表情图像数据,通过构建和训练改进后的ResNet34神经网络进行人脸表情分类;其次,为了使算法具备处理多人脸表情的能力,扩充了多人脸表情图像数据到数据集中,通过引入OpenCV分类器在自动标定和裁剪出尽可能多的人脸图像,输出每个人脸的表情分类,并将每个观众的表情分类结果映射到原多人脸照片上,以直观地展示观众的情感反应;最后,引入了统计方法,以百分比来量化图片中每种表情的分布情况,为节目效果评估提供客观指标。本文方法应用于多个情节爆发点时,可以形成时序性的节目整体效果评价,从而为文娱节目的效果评价和改进提供了一种新的思路,具有潜在应用前景。The purpose of this study is to explore an evaluation method based on expression recognition for the effect of entertainment programs.Firstly,according to the abundant expression image data in the Emotion-Domestic dataset,the improved ResNet34 neural network was constructed and trained to classify facial expressions.Secondly,in order to make the algorithm have the ability to process multi-face expressions,the multi-face expression image data was expanded into the dataset,and the OpenCV classifier was introduced to automatically calibrate and crop as many face images as possible,output the expression classification of each face,and map the expression classification results of each audience to the original multi-face photos,so as to intuitively display the audience's emotional response.Finally,a statistical method was introduced to quantify the distribution of each expression in the picture as a percentage,which provided an objective index for the evaluation of the program effect.the method can be extended to multiple points of the complete program for multi-point effect analysis,provides a new idea for the conception and production,and has potential application prospects.

关 键 词：表情识别 文娱节目 效果评价 ResNet34 OPENCV 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]