社会科学视域下的计算机视觉研究:历史、理论与方法  

Review of Computer Vision Research in the Social Science Perspective:Theory,Methods,and Prospects

作　　者：张伦[1] 樊嘉 吴晔 ZHANG Lun;FAN Jia;WU Ye(School of Art and Media,Beijing Normal University,Beijing 100875,China;Department of Media and Communication,City University of Hong Kong,Hong Kong 999077,China;School of Journalism and Communication,Beijing Normal University,Beijing 100875,China)

机构地区：[1]北京师范大学艺术与传媒学院,北京100875 [2]香港城市大学媒体与传播系,中国香港999077 [3]北京师范大学新闻传播学院,北京100875

出　　处：《新媒体与网络》2024年第2期24-39,共16页NEW MEDIA AND NETWORK

摘　　要：计算机视觉技术发展迅速,为社会科学研究者挖掘视觉数据、研究人类行为、发现新的理论“绿洲”提供了新的路径。通过系统梳理当前与计算机视觉技术相关的社会科学文献,分析总结该领域的发轫历史、发展趋势、关注议题、相关理论概念以及研究方法,发现计算机视觉研究主要涉及的理论包括视觉效果相关理论、自我认知相关理论、社会交往相关理论和心理学相关理论四类;研究议题包括政治学、媒体传播、群体行为、区域与人口研究四大方向。在方法上,社交媒体是构成计算机视觉研究最主要的数据来源,主要分析工具包括开源工具、商业平台应用程序接口(API)和研究者自己开发的工具,涵盖12种视觉分析技术。在此基础上,从计算机视觉研究为中国社会科学跨学科研究带来的新契机、计算机视觉研究的理论困境、视觉变量的测量信度与效度以及视频数据获取和使用的伦理问题四个层面,对社会科学视域下的计算机视觉研究进行了反思。The rapid advancement of computer vision technology has opened new avenues for social science researchers to extract insights from visual data,study human behavior,and discover novel theoretical“oases”.This paper aims to systematically analyze computer vision research from the perspective of social sciences,focusing on historical,theoretical,and methodological aspects.By combing current literature related to computer vision technology in social sciences,it analyzes and summarizes the origins and development trends,key topics and relevant theoretical concepts,and research methods in this field.To achieve the research objectives,this paper uses keyword research,literature retrieval and selection,and manual coding to ensure the comprehensiveness and reliability of the literature review.Specifically,on the basis of the identified core keywords,we expanded the keyword pool by using the Snowball Method through citation networks,and then searched on three literature retrieval platforms(Web of Science,Scopus,and Google Scholar),selecting 75 relevant papers from the past two decades.We then followed the basic norms of content analysis and coded the sample literature manually.Based on this,we summarized the theoretical concepts,research purposes,specific research directions,and research methods.The study reveals that the integration of computer vision technology with social sciences,as an emerging academic field,holds potential to supplement and innovate existing social science theories,thereby addressing significant research questions.Like other social science research,the core focus of this field remains closely related to human issues-both as social members and technology users.The primary theories involved in computer vision research include visual effect,self-cognition,social interaction,and psychological-related theories.The research topics encompass politics,media communication,group behavior,and regional and population studies.In terms of methodology,social media constitutes the main data source for computer vi

关 键 词：计算机视觉 社会科学研究 视觉效果 自我认知 社会交往 社交媒体 视觉变量测量 开源工具 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]