The Use of Computer Vision to Analyze Brand-Related User Generated Image Content

With the increasing popularity of visual-oriented social media platforms, the prevalence of visual brand-related User Generated Content (UGC) have increased. Monitoring such content is important as this visual brand-related UGC can have a large influence on a brand's image and hence provides useful opportunities to observe brand performance (e.g., monitoring trends and consumer segments). The current research discusses the application of computer vision for marketing practitioners and researchers and examines the usability of three different pre-trained ready-to-use computer vision models (i.e., YOLOV2, Google Cloud Vision, and Clarifai) to analyze visual brand-related UGC automatically. A 3-step approach was adopted in which 1) a database of 21,738 Instagram pictures related to 24 different brands was constructed, 2) the images were processed by the three different computer vision models, and 3) a label evaluation procedure was conducted with a sample of the labels (object names) outputted by the models. The results of the label evaluation procedure are quantitatively assessed and complemented with four concrete examples of how the output of computer vision can be used to analyze visual brand-related UGC. Results show that computer vision can yield various marketing insights. Moreover, we found that the three tested computer vision models differ in applicability. Google Cloud Vision is more accurate in object detection, whereas Clarifai provides more useful labels to interpret the portrayal of a brand. YOLOV2 did not prove to be useful to analyze visual brand-related UGC. Results and implications of the findings for marketers and marketing scholars will be discussed.