CrowdColor: Crowdsourcing Color Perceptions Using Mobile Devices

Providing accurate color information to online shopping customers is important for their purchase decisions. However, due to the multiple imaging processes that product photos undergo, end-users often experience a color mismatch between the color of the photo online and the product received. Therefore, we use a crowdsourcing approach to generate what we term CrowdColor, which is the collective color reported by individuals using a mobile color picker. CrowdColor serves as a color review application from the customers' perspectives in the form of a color palette that represents the product color. We perform controlled experiments to evaluate the accuracy of CrowdColor and to understand how the effects of the device and lighting conditions may influence the crowd's color perception and input tasks. The quantitative results reveal that CrowdColor achieves high accuracy and is positively rated overall. Based on experimental analyses, we present design guidelines for crowdsourcing color perception tasks.