Code and data on the categorization of soft-drink bottles using image silhouettes

Ana M Arboleda,Carlos Arce-Lopera

doi:10.1016/j.dib.2018.12.068

Abstract

This data article includes the visual stimuli used to test the categorization of a set of soft drink bottle silhouettes. Additionally, subjects’ perceptual categorization was associated with each visual stimuli. The silhouette of the soft drink bottles was characterized by calculating the most common object shape measurements such as width, height and area and combining them with more complex and specific quantitative shape measurements such as the principal moment statistics. Finally, this data article includes the code for extracting these shape characteristics from image silhouettes. For interpretation and discussion, please see the original article entitled “Quantitative analysis of product categorization in soft drinks using bottle silhouettes” (Arboled and Arce-Lopera, 2015) [1].

Highlights

Article history: Received 26 June 2018 Received in revised form 14 December 2018 Accepted 20 December 2018 Available online 24 December 2018 abstract. This data article includes the visual stimuli used to test the categorization of a set of soft drink bottle silhouettes
The silhouette of the soft drink bottles was characterized by calculating the most common object shape measurements such as width, height and area and combining them with more complex and specific quantitative shape measurements such as the principal moment statistics. This data article includes the code for extracting these shape characteristics from image silhouettes
Subjects were presented with the bottle silhouette images and asked to categorize them into one of seven possible options; namely, soda, fruit juice, water, tea, sports drink, flavored water and malt

Summary

Stimuli

The front side of 52 different personal-sized bottles (less than 600 ml) were photographed using a digital camera (Canon EOSD7) in a light-controlled environment. A MATLAB code was implemented to take a silhouette image as an input and measure physical properties that can be related to their shape. To measure physical characteristics that were directly related to shape of the body, we transformed the side of the bottle to a distribution and calculated four statistical features: Body Mean (BM), Body Variance (BV), Body Skewness (BS), and Body Kurtosis (BK). To achieve this shape encoding, only the Body of the silhouette was considered. The code was written in the MATLAB programming language (see ShapeMeasure.m file)

Shape data results

Psychophysical experiment results