Author(s)

YaWei He^*, ZiJing Pan, LinYi Bao, YanXi Zhu, ChiYue Zhang, ZhouChu Zhang

Affiliation(s)

School of Media and Communication, Shanghai Jiao Tong University, Shanghai 200240, China.

Corresponding Author

YaWei He

ABSTRACT

To extend the scope of computational communication in an era dominated by visual media, this study explores effective methods for analyzing video content at scale, moving beyond a traditional focus on textual data. Using a large dataset of food-related videos from the popular platform Bilibili, we developed and trained a custom deep learning model to automatically identify, extract, and quantify the presence of core visual elements within each video frame. A systematic correlation analysis was conducted to examine the relationship between these extracted visual features—specifically the face appearance rate, food appearance rate, and overall image brightness—and composite measures of the videos' communication effects. Our statistical analysis reveals that both a higher face appearance rate and greater image brightness are significant positive predictors of communication effectiveness. In contrast, and counter to common assumptions, a higher frequency of shots featuring only food was found to negatively impact the video's overall performance. These findings suggest that effective video communication relies on emotional connection rather than mere content display; facial presence significantly drives deep engagement, likely through social relationships, while the visibility of food itself negatively impact audience response, highlighting a preference for cultural context and human narratives. Furthermore, metadata such as expressive titles and channel fan count, along with release time and duration, also critically shape dissemination success. This research not only offers valuable empirical guidance for content creators but also demonstrates a replicable and cost-effective computational paradigm for large-scale video content analysis.

KEYWORDS

Artificial intelligence; Deep learning; Computational communication; Communication effect; Visual features

CITE THIS PAPER

YaWei He, ZiJing Pan, LinYi Bao, YanXi Zhu, ChiYue Zhang, ZhouChu Zhang. Deep learning driven analysis of the food-related video communication effect: investigating visual feature impacts. Journal of Computer Science and Electrical Engineering. 2025, 7(7): 18-24. DOI: https://doi.org/10.61784/jcsee3094.

REFERENCES

[1] Xiang Y, Chae S W. Influence of perceived interactivity on continuous use intentions on the danmaku video sharing platform: Belongingness perspective. International Journal of Human–Computer Interaction, 2022, 38(6): 573–593. DOI: 10.1080/10447318.2021.1952803.

[2] Song S, Zhao Y C, Yao X, et al. Short video apps as a health information source: an investigation of affordances, user experience and users’ intention to continue the use of TikTok. Internet Research, 2021, 31(6): 2120–2142. DOI: 10.1108/intr-10-2020-0593.

[3] Matei S A, Kee K F. Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery. Computational Communication Research, 2019, 9(4): e1304. https://tinyurl.com/4ekscyv4.

[4] Trilling D, Araujo T, Kroon A, et al. Computational Communication Science in a Digital Society. Communication Research into the Digital Society Fundam, 2024: 247. DOI: 10.5117/9789048560592.

[5] Joo J, Bucy E P, Seidel C. Computational communication science| automated coding of televised leader displays: detecting nonverbal political behavior with computer vision and deep learning. International Journal of Communication, 2019, 13: 23. https://tinyurl.com/37rcjsav.

[6] Wu P Y, Mebane Jr W R. Marmot: A deep learning framework for constructing multimodal representations for vision-and-language tasks. Computational Communication Research, 2022, 4(1). DOI: 10.5117/CCR2022.1.008.WU.

[7] Jürgens P, Meltzer C E, Scharkow M. Age and gender representation on German TV: A longitudinal computational analysis. Computational Communication Research, 2022, 4(1). DOI: 10.5117/CCR2022.1.005.JURG.

[8] Joo J, Steinert-Threlkeld Z C. Image as data: Automated content analysis for visual presentations of political actors and events. Computational Communication Research, 2022, 4(1). DOI: 10.5117/CCR2022.1.001.JOO.

[9]  Zhong Z, Zhou J, Su S. Computational Communication Studies in Chinese and English: Issues, Methods and Theories. Global Journal of Media Studies, 2023, 10(1): 56–76. https://tinyurl.com/yv2jyxpr.

[10]  Sun, S., & Liu, Z. Issue Visibility and Gatekeeping Actors in Weibo Trending Searches: A Computational Communication Analysis Based on Public Opinion Events. Journalism & Communication, 2024(5): 30–36+43. https://tinyurl.com/ycyfx7u3.

[11] Sun J. A study on the influencing factors of the dissemination effect of popular science journals on WeChat: a computational analysis based on 4445 tweets. Chinese Journal of Scientific and Technical Periodicals, 2023, 34(11): 1511–1520. DOI: 10.11946/cjstp.202308080600.

[12] Minghua X, Liqun Y, Ziyao W. Hidden Concurrency and Selective Incitement: The Generation Logic and Operational Mechanism of Western Computational Propaganda against China. Contemporary Communication, 2024(3): 55–60. https://tinyurl.com/5arwhdtk.

[13] Qing Y, Yu L. The social support seeking among Depressed patients on social media——Based on the online community of Depression in Weibo. Journalism and Mass Communication, 2022, (6): 45–56+64. DOI: 10.15897/j.cnki.cn51-1046/g2.20220620.004.

[14] Ahmadi I, Waltenrath A, Janze, C. Congruency and users’ sharing on social media platforms: a novel approach for analyzing content. Journal of Advertising, 2023, 52(3): 369–386. DOI: 10.1080/00913367.2022.205568.

[15]  Li H, Wu Y, Yuan F, et al. Innovative Paths of Computational Communication: A Systematic Analysis of Theoretical Framework, Research Methods and Application Practices——A Review of Computational Communication Research in 2024. Journal of Education and Media Studies, 2025(1): 32–39. DOI: 10.19400/j.cnki.cn10-1407/g2.2025.01.010.

[16] Van Atteveldt W, Peng TQ. When communication meets computation: Opportunities, challenges, and pitfalls in computational communication science. Communication Methods and Measures, 2018, 12(2-3): 81–92.

[17] Yu G, Xu W. Multimodal Fusion: Efficiency Improvement and Research Model of Media Communication. Media Observer, 2021(12): 14–20. DOI: 10.19480/j.cnki.cmgc.2021.12.004.

[18] Huang Y, Chen C. Automation of visual communication and the aesthetic construction of national image: A computational aesthetics study based on Twitter social robots. Modern Communication (Journal of Communication University of China), 2023, 45(8): 96–104. DOI: 10.19997/j.cnki.xdcb.2023.08.016.

[19] Guan L, Zhou B. Application of computer vision technology in news communication research. Contemporary Communication, 2022(3): 20–26. https://tinyurl.com/yc3vh8zw.

[20] Wu Y, Fan J, Zhang L. The Personalization Effect of Mainstream Media in the Context of ShortVideo--Content Analysis of the “Anchor Talking Broadcast” Programme. Journal of Xi'an Jiaotong University (Social Sciences), 2021, 41(2): 131–139. DOI: 10.15896/j.xjtuskxb.202102015.

[21] Xue P. Interactive Mechanism of Online Mukbang Community on Bilibili Video Website. Master’s thesis, 2020.

[22] Rodriguez L, Dimitrova D V. The levels of visual framing. Journal of Visual Literacy, 2011, 30(1): 48–65. DOI: 10.1080/23796529.2011.11674684.

[23] Wang J, Chan K C, Loy C C. Exploring clip for assessing the look and feel of images. In Proceedings of the AAAI Conference on Artificial Intelligence, 2023, 37(2): 2555–2563. DOI: 10.1609/aaai.v37i2.25353.

[24] Yu W. Did the “Shannon-Weaver” Model Ever Exist?——Critical Examination of Key Historical Facts in Domestic and International Communication Theory from an Interdisciplinary Versioning Approach. Journal of Journalism & Communication Studies, 2024, 31(10): 19–37+126. https://tinyurl.com/aadxnhh9.