PUBLICATIONS

POVeye: Enhancing E-Commerce Product Visualization by Providing Realistic Image Based Point-of-View

Author: Shogo Yamashita, Adiyan Mujibiya

We present POVeye, a method to help users in capturing and creating visualization of products for extensive representation of the product’s material color and texture. POVeye achieve this by providing realistic images captured from various angles, which are positioned correctly based on the calculated geometrical centroid. As input, users simply provide a video or multiple images of the product taken by any camera from arbitrary angles, without requiring any pre-calibration. POVeye provides an interface that shows object-centric camera positions alongside with image taken from respective camera angle. Users are able to either manually browse through automatically detected camera positions, or visualize the product by automatically detected view-angle path. POVeye leverages Structure-from-Motion (SfM) approach to obtain camera-object map. Our approach is unique from other solutions by preserving realistic imaging condition. We observe that visualization of products from different angles that provide information of light reflection and refraction potentially helps users to identify materials, and further perceive quality of a product.

Copied! instagram