Color Complexity and Sharpness for Instagram Engagement Classification in University Accounts
DOI:
https://doi.org/10.22437/proca.v2i2.55582Keywords:
Instagram engagement, visual features, temporal features, machine learning classification, social media analytics, color complexityAbstract
Understanding the visual factors that drive engagement on social media platforms is essential for effective content strategy. This study investigates the classification of Instagram post engagement using interpretable visual and temporal features extracted from university social media content. We analyzed 638 Instagram posts from two university accounts (Faculty of Science and Technology and Student Executive Board of Health Sciences), extracting six visual features (colorfulness, color complexity, dominant hue, contrast, symmetry, and sharpness) and three temporal features (posting hour, day of week, and weekend indicator). A binary classification approach was employed to categorize posts into low and high engagement classes based on median likes. Four machine learning classifiers were evaluated using stratified 5-fold cross-validation. The Random Forest classifier achieved the highest performance with an F1-score of 0.714, accuracy of 69.9 percent, and AUC-ROC of 0.775. Feature importance analysis revealed that color complexity contributed 22.4 percent to the classification decision, followed by sharpness at 21.9 percent and contrast at 16.4 percent. These findings suggest that visually complex images with higher color diversity and sharp details tend to generate greater engagement on university Instagram accounts. The interpretable nature of the extracted features provides actionable insights for social media content creators in academic institutions.
References
1. Sorensen I, Furst S, Vogler D, Schafer MS. Higher education institutions on Facebook, Instagram, and Twitter: comparing Swiss universities and universities of applied sciences. Media and Communication. 2023;11(1):264-277. doi:10.17645/mac.v11i1.6069.
2. Son J, Lee D, Oh H. Digital advertising strategies for next-generation tourism marketing: differences in visual and textual features by type of social media. J Travel Tour Mark. 2021;38(7):712-727. doi:10.1080/10548408.2021.1985438.
3. Philp M, Jacobson J, Pancer E. Predicting social media engagement with computer vision: an examination of food marketing on Instagram. J Bus Res. 2022;149:736-747. doi:10.1016/j.jbusres.2022.05.078.
4. Kanuri VK, Chen Y, Sridhar SH. Scheduling content on social media: theory, evidence, and application. J Mark. 2018;82(6):89-108. doi:10.1177/0022242918805411.
5. Capriotti P, Carretón C, Castillo A. Testing the level of interactivity of institutional websites: from museums 1.0 to museums 2.0. Int J Inf Manage. 2016;36(1):97-104. doi:10.1016/j.ijinfomgt.2015.10.003.
6. Anwar A, Kanwal S, Tahir M, Saqib M, Uzair M, Rahmani MKI, et al. Enhanced ensemble deep learning approach for breast cancer diagnosis using ultrasound images. Comput Intell Neurosci. 2022;2022:7540810. doi:10.1155/2022/7540810.
7. Dai W, Mujeeb A, Erdt M, Sourin A. Soldering defect detection in automatic optical inspection. Adv Eng Inform. 2020;43:101004. doi:10.1016/j.aei.2019.101004.
8. Varga D. No-reference video quality assessment based on the temporal pooling of deep features. Neural Process Lett. 2019;50(3):2595-2608. doi:10.1007/s11063-019-10039-6.
9. Chang AYJ, Chen KH, Pan Z. Visual attributes for social media image engagement prediction. Electronics. 2022;11(9):1398. doi:10.3390/electronics11091398.
10. Kumar N, Singh JP, Dwivedi YK, Rana NP. A deep multi-modal neural network for informative Twitter content classification during emergencies. Ann Oper Res. 2022;319:791-822. doi:10.1007/s10479-020-03514-x.
11. Jia X, Song X, Hall S. Influence of visual content on social media engagement: evidence from a high-quality image analytics approach. J Retail Consum Serv. 2023;71:103185. doi:10.1016/j.jretconser.2022.103185.
12. Shao Z, Zhang L, Li X. Improving social media engagement through image quality: evidence from Instagram. J Imaging. 2023;9(2):30. doi:10.3390/jimaging9020030.
13. Kwenda S, Roodt S. Visual features and Instagram engagement: a machine learning approach. Inf Technol People. 2023;36(7):2826-2846. doi:10.1108/ITP-09-2021-0714.
14. Ramdani P, Wardhani NK, Dermawan BA, Maylawati DS. Implementation of machine learning and deep learning in predicting the number of retweets on Twitter. J Phys Conf Ser. 2021;1869(1):012073. doi:10.1088/1742-6596/1869/1/012073.
15. Prusty S, Patnaik S, Dash SK. EfficientNet: a study on building blocks and transfer learning for image classification. Sensors. 2022;22(24):9696. doi:10.3390/s22249696.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 PROCA UNJA
This work is licensed under a Creative Commons Attribution 4.0 International License.
Published with license by LPPM Universitas Jambi. This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International (CC BY 4.0 International). This license enables reusers to distribute, remix, adapt, and build upon the material in any medium or format, so long as attribution is given to the creator.
