We present Palette, a method to objectively quantize material color and airiness to provide representative description of a product in online shopping scenarios. Photos and keywords are often used to describe color, texture, and airiness of products. However consumer photos are usually taken under uncontrolled realistic imaging conditions, whereas keywords are fuzzy and highly subjective. Palette leverages active spectrophotometry approach that involves synchronized illumination to measure the reflection and transmission properties of a material as a function of wavelength. We use a Charge-Coupled Device (CCD) sensor equipped camera to capture visible light and near-infrared light intensity. We show that by analyzing the obtained light spectrum, we are able to provide a metric to represent material color and airiness. In this paper, we describe the details in principle of operation and proof-of-concept prototype implementation, as well as reporting results of our analysis using 4 types of garments. To the best of our knowledge, Palette is the first work to exploit spectrophotometry to represent garment color, texture, and airiness; as an effort to enrich user experience in online shopping.
Publications
- Shogo Yamashita, Adiyan Mujibiya, “Palette: Enhancing E-Commerce Product Description by Leveraging Spectrophotometry to Represent Garment Color and Airiness”, Proceedings of the 33rd Annual ACM Conference Extended Abstracts on Human Factors in Computing Systems, April 2015, pp.1597-1602