The Bigger Picture: A Comprehensive Review and Recommendations for Automated Image Classification in Marketing

43 Pages Posted: 28 Sep 2022 Last revised: 29 Aug 2023

See all articles by Keno Tetzlaff

Keno Tetzlaff

University of Hamburg

Jochen Hartmann

TUM School of Management,Technical University of Munich

Mark Heitmann

University of Hamburg

Date Written: August 28, 2023

Abstract

Images can say more than a thousand words. They are a crucial part of numerous marketplace interactions, like social media communication, (online) advertising, electronic commerce, testing of new products, or experimenting with new designs. How can marketing study the associated image data? There are various image classification technologies, and some have been used in marketing research. The solution space is hard to explore, and practices vary. Because of this, it is important to know how to choose a method and how to execute and report. We conduct a comprehensive,
automated review of all marketing journals included in major rankings to identify peer-reviewed image classification applications. A summary of the decisions made in these applications provides guidance for image analysis in marketing. To understand the implications of these decisions in more detail, we conduct more than 7,200 image classification experiments (i.e., unique method-dataset-hyperparameter combinations). We find that recent transformer-based architectures outperform traditional alternatives by up to 23 percentage points. Performance also strongly depends on the area of application, and we quantify how application areas and modelling decisions relate to attainable accuracy.

Keywords: deep learning, convolutional neural networks, image classification, image analytics, method comparison, image mining

Suggested Citation

Tetzlaff, Keno and Hartmann, Jochen and Heitmann, Mark, The Bigger Picture: A Comprehensive Review and Recommendations for Automated Image Classification in Marketing (August 28, 2023). Available at SSRN: https://ssrn.com/abstract=4224968 or http://dx.doi.org/10.2139/ssrn.4224968

Keno Tetzlaff (Contact Author)

University of Hamburg

Jochen Hartmann

TUM School of Management,Technical University of Munich ( email )

Arcisstrasse 21
Munchen, 80333
Germany

Mark Heitmann

University of Hamburg ( email )

Allende-Platz 1
Hamburg, 20146
Germany

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
195
Abstract Views
680
Rank
288,116
PlumX Metrics